dissenting.
This is a capital murder appeal in which the defendant, Teddy Rose, was convicted of the purposeful or knowing murder of a police officer and sentenced to death. The Court affirms the defendant’s conviction for murder, but reverses the defendant’s death sentence. I concur in the conclusion that the death sentence should be reversed, although I do so for different reasons. I would also, however, reverse defendant’s conviction for murder.
*551The Court gives us an accurate presentation of the history and evidence in this case. The defendant shot Anthony Garaffa, an Irvington police officer, in the stomach from very close range with a sawed-off shotgun that was concealed in a canvas bag. The shooting was witnessed; defendant turned himself in and confessed to it. This evidence moved strongly to a determination that defendant committed a purposeful or knowing murder.
The majority relies on the strength of the evidence to hold that the trial court did not err in refusing to charge the jury on aggravated manslaughter, and that the trial court’s error in admitting certain prejudicial evidence was harmless. Because I differ with the majority’s assessment of the prejudice inhering in the trial court’s erroneous decisions, and because I find that both the evidence adduced at trial and the structure of the statute would support a charge, not given, on aggravated manslaughter, I would find the errors to be reversible. In addition, I believe that prosecutorial misconduct during the guilt-phase summation warrants reversal of the conviction. I concur in the majority’s decision to reverse the death sentence, but would further find in this case extreme prejudice in allowing the jury in the penalty phase to consider virtually the same evidence in support of more than one aggravating factor, thereby distorting the process involved in comparing and weighing aggravating and mitigating factors. In addition, further prejudice occurred, I believe, when the jury was permitted to consider an aggravating factor in the absence of sufficient evidence to support such a finding.
These several points justify the reversal of the defendant’s murder conviction and death penalty. This reversal, I believe, is mandated under either the conventional standard of review used by the majority or the enhanced standard I proposed and have applied in capital cases. See State v. Bey (I), 112 N.J. 45 (1988) (Handler, J., concurring); State v. Bey (II), 112 N.J. 123 (1988) (Handler, J., dissenting); State v. Koedatich, 112 N.J. *552225 (1988) (Handler, J., dissenting); State v. Zola, 112 N.J. 384 (1988) (Handler, J., concurring).
I.
The Court today upholds a trial court’s refusal to charge aggravated manslaughter in a capital murder case in which the argument that the defendant did not possess the state of mind requisite for murder was the only “defense” offered. The Court acknowledges that, in insisting on a “clear indication” of aggravated manslaughter, State v. Choice, 98 N.J. 295, 299 (1985), the trial court applied the wrong standard in deciding not to charge on aggravated manslaughter, ante at 479-480. The Court further acknowledges testimony in the record that (1) defendant’s companion was “shocked” and “surprised” when the shooting occurred, (2) defendant himself was crying and nervous after the shooting occurred, (3) the defendant turned himself in, (4) the defendant told police on turning himself in that he had “reached into the sack and when he pulled it out of the sack it apparently had cocked and when it came out of the sack the gun went off,” ante at 480, (5) the defendant stated on turning himself in that he hoped the officer would live, and (6) the defendant told police that he had shot the officer reflexively because “I panicked, did not want to get caught.” The Court acknowledges that this testimony was “the only evidence concerning defendant’s motive.” Ante at 531. The Court also concedes that certain evidence adduced by the State to rebut any state of mind defense was admitted erroneously. Ante at 488. Nonetheless, the Court purports to apply State v. Crisantos (Arriagas), 102 N.J. 265, 276 (1986), and concludes that there was no rational basis for allowing the jury to consider aggravated manslaughter in this case.
I disagree. In my opinion, a rational jury confronted with the evidence described above, expunged of the erroneously admitted, prejudicial evidence adduced by the State, could not have acquitted the defendant, but could have found the defend*553ant guilty of having recklessly caused death “under circumstances manifesting extreme indifference to human life,” N.J.S.A. 2C:11-4a, rather than of shooting the officer with the “conscious object” or “practical certainty” that he would die; such a finding would have been based, in other words, not on mere sympathy unsupported by the evidence, which would have rendered it irrational, but rather on the credited testimony that defendant’s behavior both before and after the shooting reflected a regretted act of recklessness, not a knowing or purposeful murder.
A.
By refusing to charge on aggravated manslaughter the trial court left the jury with a choice: conviction of knowing or purposeful murder or acquittal. Given that the defendant had turned himself in and admitted to the act of killing the officer, there was no way a rational jury could acquit him. The court’s refusal was, in effect, a directed verdict of guilt. A rational jury might, however, have concluded on these facts that the defendant was probably telling the truth when he stated that he had shot the officer because he had panicked and feared being caught. If given the proper instruction, the jury could have considered whether this panic reflected the defendant’s “conscious object” to kill the officer (purpose), N.J.S.A. 2C:2b(1); his “practical certainty” that his conduct would kill the officer (knowledge), N.J.S.A. 2C:2-2b(2); or his “conscious disregard” of “a substantial and unjustifiable risk” that his conduct would kill the officer (recklessness), N.J.S.A. 2C:2-2b(3), in circumstances “manifesting extreme indifference to the value of human life,” N.J.S.A. 2C:ll-4a. Having been deprived of the potentially dispositive third alternative, a rational jury could have reached only one verdict: guilty of capital murder.
The Court makes much of the fact that there was testimony from the defendant, police officers, and experts that supported a conviction of knowing and purposeful murder. I agree that *554the evidence described by the Court is sufficient to support the jury’s verdict; no one, however, is challenging the sufficiency of the evidence. In my opinion, the Court misapplies Crisantos by conducting a weighing of evidence appropriate to the jury’s deliberation rather than to the ascertainment of “rational basis” required by Crisantos. That case enjoins courts to “examine the record thoroughly” “in view of the defendant’s request” to determine whether the “low threshold” of rational basis is met. 102 N.J. at 278.
Precisely how low the threshold is can be seen by comparing the testimony in this case with the evidence adduced in Crisantos and in State v. Bohannan, 206 N.J.Super. 646 (App.Div. 1986), and State v. Powell, 84 N.J. 305 (1980), cases whose results are discussed with approval in Crisantos. In Bohannan, the Appellate Division held that it was reversible error for the trial court to have refused to charge second-degree robbery. The court noted that such a charge is warranted “[i]f there exists a rational basis in the proofs to support a conviction of the lesser degree of the crime____ [E]ven very slight evidence on [the proposed theory] will compel the instruction on a lesser offense.” 206 N.J.Super. at 649, (citing State v. Powell, 84 N.J. 305 (1980)). Turning to the facts of the case, the court found such a rational basis for second degree robbery because
[d]efendant’s testimony that he was unaware of a handgun until the [other perpetrators] returned to the car a second time was direct evidence to be weighed against the circumstantial evidence that he knew that a dangerous weapon would be used. It was for the jury to determine whether he was aware of sufficient circumstances to believe or hope that an armed robbery was contemplated. [Id. at 651.]
Thus, despite strong circumstantial evidence to the contrary, where there was evidence that, if credited, would have supported the lesser-included offense — even though that evidence derived from the defendant’s self-serving testimony — the court held that the instruction should have been given. Similarly, in this case, the majority concedes that “the only evidence concerning defendant’s motive for shooting Officer Garaffa was defendant’s own statement indicating that he panicked and did *555not want to be caught with the shotgun.” Ante at 581. The jury should have been allowed to consider whether he was telling the truth about his panic and, if so, whether his panic caused defendant to act recklessly.
In Crisantos, by contrast, this Court upheld the trial court’s refusal to charge passion/provocation manslaughter. The Court cited Bohannan with approval, and cautioned that the “rational basis” standard requires slightly more than “a scintilla” of supporting evidence, but added that the “rational basis” standard “[nevertheless ... imposes a low threshold ... for permitting a charge on a lesser-included offense.” 102 N.J. at 277-78. The Court failed, however, to find a rational basis on the record before it, because “[p]assion/provocation manslaughter in this case is not only inconsistent with the defendants’ testimony; it is also inconsistent with the State’s version of the homicide and is substantiated by no testimony in the record.” Id. at 280. The Court contrasted this situation with that in State v. Powell, supra, 84 N.J. 305, in which this Court held that the lesser-included offense of manslaughter should have been charged despite the defendant’s denial of any involvement at all and his assertion of an alibi. Manslaughter should have been charged, this Court held in Powell, because “the State introduced a contradictory statement from defendant that was held to support a manslaughter instruction.” Crisantos, supra, 102 N.J. at 280-81 (citing State v. Powell, supra, 84 N.J. at 320). Thus, the Crisantos Court concluded, “[u]nlike Powell, no evidence in this record connects the alleged provocation to the homicide.” Ibid.
Read together, Crisantos, Powell, and Bohannan establish that no rational basis exists if the proposed lesser-included offense contradicts both the defense’s and the State’s theories of the case, and is otherwise wholly unsupported by the evidence (Crisantos). A rational basis must be supported by slightly more than a “scintilla” of evidence, but can be afforded by a defendant’s self-serving statements, even in the face of strong contrary evidence {Bohannan), and by evidence ad*556duced by the State — even a single contradictory statement of the defendant — that directly contradicts both the defendant’s other statements and the defense’s entire theory of the case, so long as that evidence could be read to support the lesser-included charge (Powell). Turning to the facts of this case, it is clear that there is evidence in the record that could be read to support the lesser-included charge. This case is distinguishable, therefore, from Crisantos; the question is whether that evidence amounts to slightly more than a scintilla, and thus meets the rational basis test. In light of the truly slight evidence accepted in Bohannan and Powell, as approved in Crisantos, I believe that the rational basis threshold is met in this case by the defendant’s testimony, read in light of the surrounding circumstances, that he shot the officer out of panic that he would be caught with the weapon.
It is a rational conclusion — from the companion’s shock that the shooting occurred, from the defendant’s distraught state immediately afterward, from his surrender, his confession, his statement that he panicked, and his statement that he hoped the officer would live — that the defendant, when confronted with law enforcement, simply panicked and, manifesting extreme indifference to the value of human life in his desire to escape, cocked and pulled the trigger. To the extent that the rational basis for aggravated manslaughter rests in this case on defendant’s self-serving and contradictory statements, Bohannan and Powell are authority supporting the issuance of instructions on lesser-included offenses where there is evidence to the contrary, where the defendant has made contradictory statements on the record, and even where the lesser-included offense contradicts the theory of the defense. In this case, of course, the lesser-included offense was the theory of the defense. To the extent, moreover, that the majority finds significant the fact that all of the evidence affording the rational basis was adduced by the State, ante at 480, Powell is controlling authority that the source of the evidence does not matter.
*557The majority does not challenge this interpretation of our precedent; rather, the majority responds by concluding that “panic” is irrelevant to mental state, that “in the absence of insanity or diminished capacity, a person firing a sawed-off shotgun into the abdomen of another at point-blank range necessarily is aware that ‘it is practically certain’ that such conduct will cause the victim’s death.” Ante at 484. But cf. State v. Palmer, 211 N.J.Super. 349, 352 (App.Div.1986) (endorsing trial court’s definition of “circumstances manifesting extreme indifference,” for aggravated manslaughter purposes, - as conduct “that is practically certain to kill everyone who might happen to be in the way” (emphasis added)). The majority thus concludes that what it concedes to be “the only evidence concerning defendant’s motive,” ante at 531, was somehow irrelevant to the defendant’s state of mind. The majority accomplishes this by pointing out that “panic” is not a state of mind contemplated by the Code, by ignoring the fact that “recklessness” is contemplated by the Code, and thus by avoiding the question of whether “panic” can lead to “reckless” behavior.
Panic may not be defined in the Code — neither, I should add, are other terms of common currency, such as fear, anger, or jealousy, with which a party might be expected to describe his state of mind — but in plain English “panic” means “[a] sudden, overpowering terror,” The American Heritage Dictionary of the English Language (1978 ed.), at 948, or “a sudden terror often inspired by ... a misapprehension of danger and accompanied by unreasoning or frantic efforts to secure safety,” Webster’s Third New International Dictionary (unabridged), at 1630 (emphasis added). It was the evidence — albeit scanty— of precisely such “panic,” such “terror ... inspired by ... a misapprehension of danger” — or, as Justice O’Hern put it, such an “overreactpon] to the perceived menace” — that led this Court to hold last term, in State v. Bowens, 108 N.J. 622, 640 (1987), that the trial court should have charged the jury on aggravated manslaughter. The term “overreactpon] to the *558perceived menace,” while not defined in the Code, was held relevant in Bowens to the issue of whether the defendant had acted recklessly; if this is so, how is it that the evidence in this case of similar “panic,” or “terror ... inspired by ... a misapprehension of danger” can be considered irrelevant to the identical issue? Assuming arguendo that the jury was disposed to credit the testimony that Teddy Rose, when confronted by the police, was afflicted with “[a] sudden, overpowering terror,” it is unreasonable to preclude that jury from considering the effect of that “sudden, overpowering terror” on defendant’s state of mind. It seems self-evident that one who feels a “sudden, overpowering terror” is at least equally likely — and perhaps more likely — to act “recklessly” — with conscious disregard of the potential consequences of his act — as he is to act with “practical certainty” of the consequences of his actions. Surely if the jury is to be allowed to find that the defendant’s state of mind was “knowing” despite his panic, it should be allowed to consider the possibility that his conduct was “reckless” because of it.
The majority thus inexplicably fails to consider the question most obviously raised by the evidence relating to panic: what if the defendant was thinking only of getting away, not of where or whether he was aiming nor of how close the officer was nor, for that matter, of the nature of his instrument of escape in relation to the proximity of the person who might prevent such an escape? What if, in other words, the defendant was in a state of blind — dare I say it? — panic? An instruction on aggravated manslaughter would have “invited the jury to probe defendant’s mental processes” not, as the majority insists, to speculate whether defendant’s panic “interfered ... with his capacity to be aware of the consequences of his act,” ante at 485, but to define the extent to which that panic led the defendant to disregard those consequences: purposefully, knowingly, or recklessly, the latter with conscious — not insane or incapacitated — disregard for the consequences of his actions *559in circumstances “manifesting extreme indifference to human life.”
I believe, in short, that a rational basis existed, within the meaning of Crisantos, for charging the lesser-included offense. The Court’s approach — emphasizing the evidence against defendant, rather than the evidence that could support a rational inference — distorts precedent cynically; strong evidence supporting guilt of the greater charge will always exist, or a charge on the greater offense will be unjustified and will in fact be reversible error. By holding, in other words, that “a charge of aggravated manslaughter was unwarranted unless there was a rational basis on which the jury could find that defendant did not purposely or knowingly shoot Officer Garaffa,” ante at 482, the Court has not merely misstated the law but reformulated it. Indeed, N.J.S.A. 2C:1-8e provides expressly that “[t]he court shall not charge the jury with respect to an included offense unless there is a rational basis for a verdict convicting the defendant of the included offense.” In no prior case, moreover, has this Court suggested that a charge on a lesser-included offense was unwarranted unless there was a rational basis for acquittal of the greater offense that did not derive exclusively from the existence of evidence supporting conviction of the lesser offense. This Court noted in State v. Crisantos, supra, 102 N.J. at 277 n. 10, that the formulation in N.J.S.A. 2C:1-8e differed from the proposed wording of the comparable provision of the Model Penal Code — “rational basis for a verdict acquitting the defendant of the offense charged and convicting him of the included offense” — but declined to ascribe any significance to the distinction. The reason for this seems clear; because conviction of the included offense entails acquittal of the greater, evidence affording a rational basis for convicting of an offense involving a lesser degree of culpability will necessarily afford a rational basis for acquittal of the greater charge. Thus, by omitting the language respecting acquittal of the greater charge from the statute, while retaining it in the Commentary, the legislature expressed its conclu*560sion that this language was simply extraneous to the proper focus of the rational basis test: the existence of evidence supporting conviction of the lesser-included offense.
The majority today, however, seeks in effect not only to import the omitted language into our Code, but to manipulate it to create a two-part test. Under this new formulation, a rational basis is not established unless a defendant has not only pointed to evidence that could support a conviction on the lesser charge, but also, and quite apart from this, explained why the jury should have credited that evidence by assailing the strength of the evidence of the greater charge. In my opinion this approach transforms a “low threshold” into an unattainable ceiling. Any evidence that could be read to support conviction of a lesser charge may be said to weaken the case for the greater charge to the extent that the jury credits such evidence. Because the degree to which a jury will believe and weigh evidence is essentially inscrutable, however, the question of what weight to afford evidence that could be read to support a lesser charge has until today been left to the jury; rational basis has been afforded in our prior cases by the existence of slightly more than “a scintilla” of evidence of the lesser charge, not by any assessment of its weight as against the evidence supporting conviction of the greater charge. The majority’s approach thus transforms the rational basis test by requiring courts to “hypothesize” the weight a jury would have ascribed to the evidence supporting conviction of the lesser charge. Furthermore, because all that is required to establish a rational basis is slightly more than a “scintilla” of evidence of the lesser offense, such evidence will almost always be outweighed substantially by evidence sufficient to establish guilt of the great offense beyond a reasonable doubt. To the extent, therefore, that the majority’s approach requires courts to assess the weight of the evidence of the lesser offense as against the weight of the evidence of the greater, it departs from all precedent, imposes an unrealistic standard, and usurps the function of the jury. To the extent, moreover, that the majori*561ty’s approach implies that the strength of the evidence supporting conviction of the greater charge should be evaluated independent of the evidence supporting conviction of the lesser standard, it sets an unattainable standard; if there were a rational basis for acquittal that was afforded not by the evidence supporting a lesser charge but by the weakness of the evidence of the greater charge, the defendant should never have been convicted of the greater offense in the first place. Such a “rational basis” independent of the evidence of the lesser charge would be difficult to distinguish, in other words, from a “reasonable doubt” justifying acquittal. Few, if any, defendants raising this issue will be able to meet such a standard. In my view, the only evidence relevant to the possibility of acquittal of the greater charge, for purposes of ascertaining whether “there is a rational basis for a verdict convicting the defendant of the included offense,” N.J.S.A. 2C:1-8e, is the evidence that has been the focus of the rational basis inquiry in all of our prior cases: the evidence that could be read to support conviction of a lesser charge.
The majority’s standard, rather than restating this “low threshold,” transmutes it, and seems insurmountable on its face. It is, moreover, precisely the approach that this Court rejected just last term by affirming the Appellate Division’s decision in State v. Bowens, 205 N.J.Super. 548 (App.Div.1985), aff’d, 108 N.J. 622 (1987). In that case, a majority of the Appellate Division panel reversed a murder conviction for the trial court’s failure sua sponte to charge the jury on aggravated manslaughter. The Appellate Division reviewed the record testimony and, relying chiefly on “defendant’s own testimony,” 205 N.J.Super. at 551, found “[a] version of the evidence which would have supported a jury verdict of aggravated or reckless manslaughter____” The Appellate Division dissent, like the majority’s approach in this case, emphasized the strength of the evidence of murder, and noted that defendant had “conceded in his brief that ‘there is more evidence on the record to infer a finding that defendant’s actions were either purposeful or *562knowing’ rather than reckless.” Id. at 556 (dissenting opinion). This Court rejected the dissenting approach, agreeing with the approach of the Appellate Division majority, which treated as irrelevant the strength of the evidence of murder, emphasizing instead the fact that defendant’s version afforded “sufficient evidence ... to have the jury consider whether the defendant overreacted to the perceived menace____” 108 N.J. at 640. There was no intimation that “overreact[ion] to ... perceived menace” was irrelevant because it is not defined in the Code or that defendant had to show a rational basis for not convicting apart from the evidence that could have been read to support a conviction of the lesser charge.
The majority’s formulation, requiring that defendant show a rational basis for not convicting him of the greater charge, not only requires defendants to prove a negative, but also shifts the focus of the rational-basis inquiry from its proper object: whether evidence exists that, if credited, would support the lesser-included charge. Because such evidence existed, and because no such charge was given, defendant’s conviction should be reversed.
B.
A second, perhaps more important, reason that the jury should have been allowed to consider convicting of aggravated manslaughter is that it was allowed to convict of knowing murder. The distinction between knowledge and recklessness, between “practical certainty” of a result and “conscious disregard” of “a substantial and unjustifiable risk” of a result, is a subtle one at best. Under New Jersey’s capital murder statute, a defendant can be considered death-eligible for the “knowing” infliction of “serious bodily injury resulting in death,” but as the trial court instructed the jury in this case, serious bodily injury “can be defined as bodily injury which creates substantial risk of death,” ante at 528. See N.J.S.A. 2C:11-1b. Similarly, “the risk under [both N.J.S.A. 2C:11-4a and b, aggravated and reckless manslaughter] must be ‘a substantial risk of *563death.’ ” Cannel, New Jersey Code of Criminal Justice Annotated (1987), at 259. Thus, the term on which “knowledge” is predicated in the capital murder statute, “serious bodily injury,” denotes the very idea of “substantial risk” of death on which aggravated manslaughter is predicated; the distinction between “knowing” murder and “aggravated manslaughter” turns in close cases, therefore, on the difference between “practical certainty” that one is inflicting injury with a substantial risk of death and a conscious disregard of a substantial risk of death that manifests “extreme indifference to the value of human life.” The Commentary to the proposed Code revision of 1971, which introduced the concepts ultimately codified as “knowing” murder and “aggravated manslaughter,” made clear the fineness of the distinction between them by suggesting that it did not exist:
[Tjhere is a kind of reckless homicide that cannot be distinguished ... from homicides committed knowingly. Recklessness presupposes an awareness of the creation of substantial homicidal risk____ Since risk, however, is a matter of degree and the motives for risk creation may be infinite in variation, some formula is needed to identify the case where recklessness should be assimilated to knowledge. The conception employed is that of extreme indifference to the value of human life. The significance of purpose or knowledge is that ... it demonstrates precisely such indifference. Whether recklessness is so extreme that it demonstrates similar indifference is not a question that, in our view, can be further clarified; it must be left directly to the trier of the facts.
[New Jersey Penal Code, Volume II: Commentary, Pinal Report of the New Jersey Criminal Law Revision Commission, 1971 (quoted in Cannel, New Jersey Criminal Code Annotated (1987), at 245-46) (emphasis added).]
The “indifference” signified by “knowledge” is, therefore, a species of “extreme indifference” to “substantial homicidal risk”; a “conscious disregard” of such “substantial homicidal risk” in other circumstances “manifesting extreme indifference to the value of human life,” the Commentary held, raises reckless manslaughter to the equivalent of knowing murder.1 *564Thus, the Commentary recognizes the propinquity — indeed, it urges the inseparability — of species of recklessness “manifesting extreme indifference to the value of human life” and “knowing” murder; “extreme indifference” is the formula by which “recklessness” is “assimilated to knowledge.”
Perhaps in recognition of the minuteness of any distinction between aggravated manslaughter and knowing murder, aggravated manslaughter was omitted from the original enactment of the Code revision. In 1979, however, the aggravated manslaughter provision was added — but only as a separate and, in proper cases, a lesser-included offense — in the language proposed in 1971 as the equivalent of knowing murder; it applied, in other words, “when the actor other than purposely or knowingly causes death under circumstances manifesting extreme indifference to human life.” L.1979, c. 178.2 By further amendment in 1981, the language of the provision was amended to make clear that “extreme indifference” established a species of “recklessness,” the precise term discussed in the 1971 Commentary, rather than “negligence.” L.1981, c. 290. Cannel, supra, at 258-59. Thus, the Commentary’s discussion of the distinction between knowledge and recklessness that manifests extreme indifference speaks directly to the terms of the current statute. Cf. State v. Palmer, supra, 211 N.J.Super. at 352 (endorsing a definition of “circumstances manifesting extreme indifference” as conduct “practically certain to kill anyone who might happen to be in the way”).
Aggravated manslaughter is not a class of murder, as the Code revision intended, but its formulation in terms identical to *565those discussed in the Commentary, and the Commentary’s conclusion that those terms rendered it indistinguishable from knowing murder, should give this Court pause. This is particularly so when one considers the consequences of the distinction under the current Code. When the original Code revision was proposed, the Commentary took pains to point out that “[i]t is only purposeful killings which subject the defendant to capital punishment.” 1971 Commentary (quoted in Cannel, supra, at 245). Thus, when the commentators equated what we know as aggravated manslaughter with knowing murder, any proffered distinction between them could have made a difference in degree only; this was the situation from 1979, when aggravated manslaughter was adopted, until 1982. When the Legislature included “knowing” murder in the 1982 amendment reinstituting capital punishment, however, it transformed a difference in degree — and a tenuous one at that — into a difference in kind. Last year, this Court upheld the constitutionality of this scheme. See State v. Ramseur, 106 N.J. 123, 194-95 (1987); but see id. at 389-90 (Handler, J., dissenting).
Today this Court goes even further, holding that the distinction between “knowing” murder and “aggravated manslaughter” — a distinction the existence of which the Code Commentary of 1971 denies utterly, but a distinction that can now mean the difference between life and death — is great enough to justify charging the jury on knowing murder, for which the defendant is death-eligible, while refusing to charge the jury on aggravated manslaughter, for which the defendant is not death-eligible. I cannot agree. In my opinion, the distinction between a knowing infliction (an infliction with practical certainty) of serious bodily injury (a substantial risk of death) and a conscious disregard of a substantial risk of death that manifests extreme indifference to the value of human life is, as the Commentary suggests, one for juries to draw deliberatively, not one for trial judges to draw preemptively, particularly when that distinction is the defendant’s only argument.
*566At a minimum, this analysis suggests that in capital cases in which the jury is charged on “knowing” murder it should also be charged on aggravated manslaughter. To paraphrase the original Code Commentary, whether the degree of “indifference” to “substantial homicidal risk” more closely approximates “practical certainty” or “a conscious disregard” “manifesting extreme indifference to the value of human life” is not a question that can be further clarified; “it must be left directly to the trier of the facts.” Accordingly, this conviction should be reversed and the case remanded for the court’s refusal to charge on the “lesser-included offense” of aggravated manslaughter.
Beyond adopting this prophylactic measure, this Court should face the implications of the proffered analysis for this State’s capital punishment system. By enacting aggravated manslaughter as a lesser-included offense of knowing murder in terms that, according to the original Code Commentary, render the two indistinguishable, and then by including knowing murder as a capital offense, the Legislature has created a system that will of necessity function irrationally. Even where a charge is given on the lesser-included offense, the distinction between defendants convicted of knowing murder and sentenced to die, defendants convicted of knowing murder whose lives are spared, and defendants convicted, instead, of aggravated manslaughter will be difficult, if not impossible, to draw. By holding as it does today, moreover, this Court exacerbates this irrationality; there will now be verdicts of guilt of knowing murder that are unreliable because the jury was not allowed to consider convicting of the nearly identical, but non-death-eligible, crime of aggravated manslaughter. This Court is enjoined, however, by the statutory requirement of proportionality review, to attempt to compare “like” cases to ensure rationality.
The Court’s dilemma is, therefore, clear: by upholding this statutory structure, it has made an effective proportionality review more necessary even as it has rendered it next to impossible. In addition, the charging discretion left to prosecu*567tors by this statutory structure, particularly in view of this Court’s unwillingness to set guidelines to channel prosecutorial discretion, State v. Koedatich, 112 N.J. 225 (1988), seems to me certain to result in irreconcilable charging decisions.
The Legislature’s enactment of aggravated manslaughter as a lesser-included offense, coupled with its inclusion of knowing murder as a capital offense, bespeaks inadvertence rather than deliberation, an inadvertence reflected in the absence of any legislative history explaining why aggravated manslaughter is now an offense distinguishable from knowing murder, and in the statement of the capital murder legislation’s sponsor that this State’s capital murder statute is “not as broad” as statutes in some other states, in part because defendants are death-eligible only if “found guilty ... of first degree murder, willful, premeditated murder,” a degree of murder closely corresponding only with a “purposeful” state of mind. Capital Punishment Act: Hearings on S.112 Before the Senate Judiciary Committee (1982) at 1. Because I see little distinction between aggravated manslaughter and knowing murder, because my belief is unshaken that the inclusion of knowing murder as a capital offense has contributed to the creation of an unconstitutionally overbroad class of capital defendants, and because I am convinced that the Legislature’s adoption of aggravated manslaughter as, in effect, a lesser-included offense, coupled with its inclusion of knowing murder as a capital offense, has created a system that can function only in an arbitrary and capricious fashion, I am constrained to renew my objections to the constitutionality of the capital murder statute. See State v. Bey (II), 112 N.J. 123 (1988) (Handler, J., .dissenting); State v. Ramseur, supra, 106 N.J. 343 (Handler, J., dissenting).
II.
The Court’s stringency in deciding that there was no rational basis for charging aggravated manslaughter stands in sharp contrast to its assessment of other evidentiary rulings and to *568its treatment of prosecutorial misconduct during the guilt phase. The Court upholds and confirms harmless evidentiary rulings whose relevance is admittedly attenuated and whose capacity for prejudice is both flagrant and undisputed. In addition, the Court glosses over a glaring instance of prosecutorial misconduct during the guilt phase of the trial. In my opinion, both issues warrant reversal.
A.
The Court strains to uphold as relevant to the defendant’s state of mind the admission of testimony concerning defendant’s brandishing of the gun on another occasion. The Court concedes that this testimony was inadmissible “to prove that on August 8, 1984, defendant possessed the shotgun with the purpose to use it unlawfully.” Ante at 488. Nonetheless, the Court upholds the admission of the testimony as “somewhat relevant to the issue of defendant’s state of mind at the time of shooting,” ante at 488, and points, as support for this position, to the fact that defendant requested, “at the close of the guilt phase ... a charge of aggravated manslaughter.” Ante at 489.
The Court cannot have it both ways. Surely if the playground conversation is admissible as “somewhat relevant” to the defendant’s state of mind, the concentration of other evidence discussed above — the surrender, confession, remorse, etc. —affords a “rational basis” for refuting what this “somewhat relevant” evidence is allowed to prove. Furthermore, if the evidence of the playground incident, which involved defendant’s antipathy toward a black man, is relevant at all to defendant’s shooting of the white policeman — and its relevance is, in my view, so remote as to require exclusion — that relevance lies in revealing a defendant ready to use his weapon without regard for the identity of his target and therefore with reckless disregard for the value of human life — the very definition, it seems to me, of aggravated manslaughter. See State v. Palm*569er, supra, 211 N.J.Super. at 352. If the Court insists that this evidence was properly admitted, it must consider interpretations of this evidence that might support a charge of aggravated manslaughter. State v. Powell, supra, 84 N.J 305.
This reading of relevance is buttressed by the admission of testimony that the defendant had purchased the shotgun because “he was having some problems over in Jersey with some niggers.” The Court concludes that this testimony, like the playground testimony, was “somewhat probative of the question of whether defendant’s shooting ... was purposeful or accidental,” then holds that its admission, if erroneous, was harmless. Ante at 488-489. If this evidence is “somewhat probative” on whether the conduct was purposeful or accidental, it was equally probative, given that defendant’s problem was with “some niggers” and his victim was a white policeman, on whether that state of mind was extreme indifference to the value of human life; if it was properly admitted, the jury should have been given the option to consider its relevance in this respect.
In any event, it seems to me clear that the capacity of this evidence for prejudice, for convincing the jury that this was, as the prosecutor put it, just a “bad guy” who, in addition to killing a white cop, hated blacks, outweighed any relevance it might have to the defendant’s state of mind. Evid.R. 4. Its effect will in all likelihood, moreover, continue to be felt, as the record of the guilt phase is customarily entered into evidence at resentencing. The capacity of this evidence that the defendant is a racial bigot to corrupt the moral balancing of sentencing, to function, in other words, as a non-statutory aggravating factor, seems to me incalculable. At a minimum, the Court should order this evidence expunged from the record introduced into evidence at the resentencing hearing.
The Court justifies its search for “the possible grounds” for upholding the admissibility of the evidence by noting that such a search “underscores the necessity for according some mea*570sure of discretion to trial courts” where the evidence “may be both material and inflammatory.” Ante at 489. This degree of deference is inexplicable in a capital case, particularly where, as here, the rulings have nothing to do with demeanor or other factors within the unique purview of the trial court. The majority’s repeated protestations that its review of the record in capital cases is independent and searching, e.g., State v. Bey (II), supra, 112 N.J. 123; State v. Koedatich, supra, 112 N.J. 225, take on an inadvertent irony in this case, where the Court searches the record scrupulously to find “the possible grounds” to hold this prejudicial evidence admissible as relevant to state of mind, while holding that the only record evidence relating to the defendant’s motive was irrelevant to his state of mind and discounting other evidence that could be read to support a rational basis for aggravated manslaughter.
This evidence, in short, should not have been admitted. Even if the Court is correct to uphold its admission, however, it can be read with other evidence in the case to afford a rational basis for a charge of aggravated manslaughter. In either case, in my judgment, the conviction should be reversed.
B.
The prosecutor’s numerous excesses during the penalty phase are treated at length by the majority. Ante at 508-524. The majority gives short shrift — indeed, no shrift — however, to an instance of improper prosecutorial comment during the guilt phase that corrupted the entire working of the statute. During his closing argument, the prosecutor drew the following comparison between defendant and the victim:
This photograph of Teddy Rose, this is the man who took the life of an honest hard working law abiding citizen who was on the streets to protect you and I. This is the difference between you. The police officer is a different, this is a different person than you, Teddy Rose. This Officer Garaffa is a person like you and I who does his job day in and day out to raise a family. There’s something else about him though that makes this the serious case that it is here and we talked about this in jury selection.
*571He, Officer Garaffa, stands between you and this man, the likes of this man. Take him away, we’re nowhere, we’re not in New Jersey, we’re not in Essex County, we’re not in the free law abiding country. You have a serious responsibility as I stated.
It should go without saying that Teddy Rose is no more or less culpable for knowing or purposeful murder because his victim was a person “who does his job day in and day out to raise a family.” Defendant either committed knowing or purposeful murder or aggravated manslaughter or he did not; the victim’s status in the community is absolutely irrelevant to the degree of culpability during the guilt phase.
The prosecutor went further, however, arguing that the fact that the victim was a policeman “makes this the serious case that it is here.” This comment is as egregious as any that could occur during a capital case, for its effect is to corrupt the structure of the entire statute. The prosecutor was correct that the fact that the victim was a police officer “makes this the serious case that it is; ” it does so, however, by virtue of aggravating factors c(4)(h) and, in this case, c(4)(f), which are properly before the jury during the penalty phase. By injecting a penalty phase aggravating factor into the guilt phase deliberation, the prosecutor both ignored and frustrated a principal purpose of conducting a bifurcated proceeding in a capital case: to prevent the guilt phase deliberation from becoming infected with circumstances that might aggravate the crime as far as sentencing is concerned, but that bear no relevance to and substantial capacity for prejudice during the guilt phase deliberation.
Justice Clifford, concurring, agrees that this statement constituted misconduct, but concludes that it was rendered harmless beyond a reasonable doubt by “the ready acknowledgement of defense counsel in his summation that his client was guilty of knowing or purposeful murder.” Ante at 547. Justice Clifford quotes at length from this summation, ante at 548-549, as an indication that nothing the prosecutor said could have made things worse for defendant; this is all too true with *572respect to the jury’s deliberation of guilt. I agree, moreover, with Justice Clifford that both the prosecutor’s comment and the defense counsel’s summation must be seen in context. I differ, however, from Justice Clifford in my assessment of that context, and in the weight that I attach to defense counsel’s summation.
Prior to delivering his summation, defense counsel had requested an instruction on aggravated manslaughter; the rejection of this request, despite the evidence, discussed supra, that could have supported the charge, effectively directed a verdict of guilt. Justice Clifford thus makes too much of the defense counsel’s “ready acknowledgement” of guilt; what else was defense counsel to do? Rather than substantiating the harmlessness of the prosecutor’s remarks, defense counsel’s summation, read in context, speaks eloquently of how harmful was the trial court’s refusal to charge on the lesser-included offense. The jury’s deliberations were co-opted; there was, as defense counsel recognized, nothing for the jury to decide:
You know that a finding of guilty of purposeful knowing murder forces us to go to a penalty phase where the alternatives are either life imprisonment ... or the death penalty.
It’s difficult for him to admit that, it’s difficult for us to concede it and perhaps for you to accept it, to accept those will be the only options available to you____
Viewed in this context, the prosecutor’s guilt-phase remarks are the more egregious, for they reveal that his corruption of the statutory structure was as gratuitous as it was calculated. The only explanation for the prosecutor’s behavior, after defense counsel’s summation, is not that he was in zealous pursuit of a conviction — that much had been conceded — but that he was using the guilt-phase deliberation to condition the jury’s penalty-phase sentence. This behavior is a clear violation of due process, and should be, like the wrongful deprivation of a peremptory challenge, reversible without regard to a showing of actual prejudice.
In State v. Ramseur, supra, 106 N.J. at 322-23, we admonished prosecutors that we would not hesitate to reverse capital convictions based on prosecutorial misconduct where that con*573duct is “so egregious that it deprived defendant of a fair trial.” This is such a case; the “fairness” of a trial depends not on the culpability of a defendant, for it seems obvious that the most guilty defendant can have a trial that was unfair; rather, the fairness of a trial must rest, particularly in capital cases, on the integrity of the procedures that safeguarded the trial. The instance noted demonstrates that the misconduct in this case did not begin with the penalty phase, but infected the entire proceeding and, indeed, made a mockery of the statutory scheme by insisting that jurors consider what made the crime aggravated even as they deliberated whether it was a crime at all. As Justice Clifford put it in State v. Koedatich, supra, 112 N.J. at 342: “I see little hope of avoiding repetition of the deprivation of a fundamental constitutional right to a fair trial if we do no more than ‘reiterate our warning ... ’ that dire consequences may flow from [prosecutorial] violations____” Indeed, by virtually ignoring this instance of misconduct, the majority sends an entirely wrong signal. The defendant’s conviction must be reversed.
III.
In this case, I find that serious errors have affected the weighing process that is critical in a jury’s determination of whether the defendant lives or dies. The majority concludes quite soundly that there must be a reversal of the death sentences for errors infecting the jury’s deliberation of penalty. I agree. In addition, I believe the weighing process was irretrievably distorted by improperly multiplying aggravating factors, and, further, by permitting the jury to consider and deliberate on an aggravating factor for which there was simply no evidential support.
A.
I have expressed, in prior cases, my view that the reliability of capital sentences is undermined by the potential for ostensi*574bly distinct aggravating factors to be based on identical evidence. See State v. Ramseur, supra, 106 N.J. at 392 n. 21; State v. Bey (II), supra, 112 N.J. at 155-163. The Court has rejected this argument, acknowledging the potential for prejudice but holding that so long as the jury is made aware “that it is considering the same facts more than once, and [is made] cognizant that the same facts are being used to prove more than one aggravating factor,” id. at 177, double counting is permissible.
The facts of this case illustrate the problem. The jury was instructed that it could find, among other factors, factor c(4)(f) (that “the murder was committed for the purpose of escaping detection, apprehension,” etc.) and factor c(4)(h) (“The defendant murdered a public servant ... while the victim was engaged in the performance of his official duties____”). Thus, the jury was allowed to find that the murder was aggravated both because defendant murdered to escape detection, and because he murdered the person whose duty it was to detect him. Because the victim’s “performance of his official duties” was, in fact, the “detection” from which defendant sought to escape, the two aggravating factors are not only overlapping but nearly identical; given the Court’s acknowledgement of the potential prejudice, one wonders what the point is of allowing the jury even to consider both factors.
While it is true, moreover, that there is “a series of cases from other states” allowing the submission of overlapping aggravating factors, it is not true, as the majority’s analysis implies, that California is the only state to prohibit the submission of overlapping factors. Indeed, the case law on this point seems, if anything, evenly divided. While the majority cites cases from Georgia, Ohio, Maryland, and North Carolina that allow for the submission of overlapping factors, it is worth noting that, in addition to California, Alabama, Florida, and Nebraska prohibit the submission of overlapping factors. See, e.g., Cook v. State, 369 So.2d 1251, 1256 (Ala.1979); Provence v. State, 337 So.2d 783, 786 (Fla.1976) (prohibiting submission *575of both murder committed in course of robbery and murder committed for pecuniary gain because “both subsections refer to the same aspect of the defendant’s crime____ [O]ne who commits a capital crime in the course of a robbery will always begin with two aggravating circumstances against him____”); State v. Rust, 197 Neb. 528, 250 N.W.2d 867 (1977). Furthermore, one justice from another state that has allowed the submission of overlapping aggravating factors has expressed misgivings. In Wiley v. State, 484 So.2d 339, 351-52 (Miss.), cert. denied, 479 U.S. 906, 107 S.Ct. 304, 93 L.Ed.2d 278, reh. denied, 479 U.S. 999, 107 S.Ct. 604, 93 L.Ed.2d 604 (1986), Justice Robertson, who had endorsed the Mississippi Court’s allowance of overlapping factors, departed from this view, stating:
[N]o one doubts that the side with the largest number of “circumstances” has a practical advantage before the sentencing jury____ In the end, the fallacy of our rule is its failure to recognize that murders are aggravated by a defendant’s conduct, not by statutory language____ A single legally indivisible act of the defendant may rationally aggravate a murder but once. [Id. at 358 (Robertson, J., concurring).]
It seems to me clear that the inevitable result of allowing the jury to consider both factors in this case is an artificial inflation by statutory language of the aggravating circumstances of the crime. The submission of separate factors carries with it the implication that each factor that is found escalates the degree of aggravation; why else would they be separate? One simply cannot separate, however, the reason for this defendant’s conduct, the basis of one factor, from the performance of the officer’s duty, the basis of the other. The defendant acted the way he did precisely because the officer was acting in the course of his duty. The distinction between motive and result is, therefore, entirely specious in the circumstances of this case.
The Court acknowledges (if it undersells) the potential for prejudice, but holds that this can be cured by a limiting instruction to the jury that will “prevent it from giving undue weight to the number of factors.... ” Ante at 526 (quoting State v. Bey (II), supra, 112 N.J. at 177.) Try to imagine the effect of *576this instruction from the perspective of a rational juror. This juror will be instructed that he or she will be conducting a weighing process, in which the jury will be asked to enumerate certain aspects of the crime deemed by the legislature to “aggravate” it; the inevitable implication of this task is that the degree of aggravation of the crime increases with the number of these presumably separate circumstances that are found to exist. This implication is undermined where the aggravating element underlying two factors is identical. The Court recognizes this, so the juror will now be instructed, in essence: “Look. We want you to be aware that even though these two circumstances may exist, you have considered the same facts twice. You are allowed to do this, but we want you to know that you are doing it. Do not give those factors undue weight.” A rational juror, confronted with this charge, may well wonder what is meant by “undue weight.” Does undue weight consist in giving the second factor any weight at all? Giving both full weight? Giving both three-quarters the weight each would have if considered alone? Giving one full weight and the other half? Does not the fact that there are two circumstances rather than one increase the degree of aggravation?
The inadequacy of the majority’s hybrid instruction is patent. The “undue weight” is accorded automatically when the jury is allowed to find that this murder was aggravated both because defendant murdered to escape detection and because he murdered the person whose duty it was to detect him. The jury should not be allowed, on remand, to find both e(4)(f) and c(4)(h).
B.
The jury’s conduct of the weighing process was further corrupted by its consideration of an aggravating factor, c(4)(c), that was unsupported by the evidence. The majority appropriately concludes that because the sentence is being vacated, “we *577need not now definitively resolve the question whether the improper submission of an aggravating factor ... necessarily constitutes reversible error.” The Court’s gratuitous analysis suggests, however, that because the jury’s rejection of one aggravating factor “neither compels nor inhibits its determination that another factor exists,” our case of State v. Christener, 71 N.J. 55 (1976), is inapposite. I reject this dictum.
Christener stands for the proposition that it is reversible error to instruct the jury on a degree of a crime that is more serious than the evidence will support. This is so, the Court held, because “[i]t must be assumed that the jury inferred by the giving of such an instruction that the elements of that charge were present in the case.” Id. at 73. The majority has held that the aggravating circumstances in the capital-murder statute function, in narrowing the class of death-eligible defendants, as elements of the offense. State v. Ramseur, supra, 106 N.J. at 201 n. 27. Thus, to the extent that an aggravating circumstance is charged that is unsupported by the evidence, the very evil identified in Christener — an artificial inflation of the elements that the jury is entitled to assume are present— corrupts the deliberations. The majority’s proffered distinction, moreover, that “rejection of one factor neither compels nor inhibits its determination that another factor exists” is unpersuasive. Given the jury’s inference from the charge that the elements are present in the case, it seems likely that the greater the number of aggravating factors charged, the more likely the jury is to believe that at least some of those factors are present. This is precisely the kind of compulsion toward a jury’s acceptance, in the interest of compromise, of an element it might otherwise have rejected that the Christener Court expressly held requires reversal. For this reason, the majority’s dictum should be rejected.
IV.
More than anything, the Court’s result today reveals the corrosive effect of the death penalty on this State’s general *578criminal jurisprudence. Just as in State v. Koedatich the principle of a “realistic likelihood of prejudice” has been so distorted that a change of venue will now be virtually impossible to obtain, and just as in State v. Bey (II) the circumstances under which one can be said to have invoked the right to cut off questioning have been narrowed so that ambiguity is no longer sufficient, so in this case the “low threshold” of evidence required to support an instruction on a lesser-included offense has been raised considerably, the only record evidence of a defendant’s motive has been declared irrelevant to his state of mind, the scope of what undisputably inflammatory evidence is admissible as relevant to a defendant’s state of mind is broadened substantially, and the scope of allowable prosecutorial ’comment is expanded so that the integrity of the statute is undermined.
My differences with the Court in this case, as in prior cases, reflect the fact that the Court has employed a more tolerant and less strict substantive standard for determining reversibility than is demanded in the direct appeal of capital-murder convictions. See, e.g., State v. Bey (I), 112 N.J. at 93-95 (an error is reversible if it has not produced an “unjust result,” R. 2:10-1, or if it does not appear beyond a reasonable doubt that the error contributed to the jury’s guilty verdict, citing State v. Macon, 57 N.J. 325 (1971)). However, I believe that even under this substantive standard of reversible error there should be a reversal. Whether the result is “unjust” — the imposition of a death sentence — cannot be determined merely by a review of evidence. Moreover, I do not believe it can be concluded beyond a reasonable doubt that certain of the errors, discounted by the majority, did not contribute to the jury’s verdict of guilty. The majority’s result illustrates how elastic the standard for harmless error can become, and how easily courts can succumb, in capital cases, to the temptation to defer inappropriately to trial court judgments. It is for these reasons that I have advocated adoption of a clearly defined standard of review for capital cases.
*579We should, I submit, focus not on results in capital cases, but on the integrity of the procedures that produce them. Result-oriented tests such as “harmless error” call essentially for a quantification of evidence, a comparing and weighing of proper evidence with improper evidence to determine simply whether the former can satisfactorily account for and explain the jury’s determination without the latter. The assumption of this test, perhaps well-grounded for the generality of criminal appeals, is that the jury’s determination of guilt can be viewed as primarily the consideration and assessment of evidence, and only tangentially as the exercise of conscience. That assumption, I suggest, must be discarded in a capital-murder prosecution, the only proceeding in which the jury is not told that it should find the facts with no thought for the ultimate sentence.
When the State insists that a criminal pay with his life, then it must fully, not begrudgingly, acknowledge and accept the jury as the conscience of the community. The evidence in a capital case is not, then, simply weighed and quantified, but rather sorted and seen through the prism that constitutes our collective community conscience. It is this insight of the jury’s responsibility and deliberative function that impels us to adopt a more exacting standard for reversible error in capital cases. Our review should be one that answers the question of whether there was any realistic possibility that error had an adverse or detrimental influence on the deliberations of the jury. Such an effect or influence is one that can cloud the thinking, alter the feeling, or color the conscience of the jury. That alone should suffice to impugn a verdict for capital murder, not whether on some scale the error itself, in whole or part, could be said to have caused the verdict.
Justice O’HERN joins in Part II of this opinion.
Chief Justice WILENTZ joins in the majority opinion on the penalty phase but has filed a separate dissenting opinion on the guilt phase.
*580Justice O’HERN joins in the majority opinion on the penalty phase, but joins in Part II of Justice HANDLER’S dissenting opinion in respect of the guilt phase.
For affirmance of convictions and reversal of sentence; remand — Justices CLIFFORD, POLLOCK, GARIBALDI and STEIN-4.
For reversal of convictions and sentences; remand — Chief Justice WILENTZ and Justices HANDLER and O’HERN-3.
There is no intimation in the Commentary that the “motives for risk," such as panic, must be codified, as the majority argues, ante at 484, to be relevant to a defendant’s state of mind; indeed, the Commentary’s conclusion that such *564“motives for risk creation may be infinite in variation” suggests that such motives, while relevant to “recklessness,” are not susceptible of codification.
The same legislation also clarified the scope of knowing murder by adding the "serious bodily injury" language. Thus, “practical certainty” was expressly made applicable to more than just death; the "serious bodily injury” language extends its applicability to injury carrying a "substantial risk of death” — the precise scope of aggravated manslaughter.