THE PEOPLE, Plaintiff and Respondent, v. DAVID LESLIE MURTISHAW, Defendant and Appellant, 29 Cal. 3d 733

Opinion TOBRINER, J. On April 9, 1978, defendant and Gregory Lufenberger, his brother-in-law, went shooting in the Mojave Desert. They chanced upon four college students who were filming a movie. Several hours later, as the students prepared to leave, defendant began to fire at them, killing three of the students and injuring the fourth. Defendant admitted the shooting but claimed diminished capacity. The jury rejected the defense and found him guilty of three counts of first degree murder, and one count of assault with intent to commit murder.

The jury found the special circumstance of multiple murder and returned a verdict imposing the death penalty. The trial court denied defendant’s motions for modification of sentence and for a new trial. Defendant’s appeal to this court is automatic. (Pen. Code, § 1239, subd. (b).) The State Public Defender, representing defendant on this appeal, presents a number of significant contentions in his behalf. We state our conclusions respecting those contentions. 1.

Defendant contends that no substantial evidence supports the verdict of first degree murder on either a theory of premeditated murder or a theory of felony murder. We explain that the evidence is clearly adequate to show a deliberate and premeditated killing.

Although the evidence that defendant killed in the course of an attempt to rob the victims of their automobile is marginal, consisting almost entirely of Lufenberger’s testimony that defendant earlier in the day said “he wanted to steal the car or something.” In view of the convincing evidence of premeditated murder, any error in submitting the issue of felony murder to the jury was not prejudicial. 2. The day after the crime sheriff’s deputies interviewed defendant at the Norwalk police station, then took him to Bakersfield where they taped a second interview.

In both interviews defendant admitted the killings, but did not admit to malice, premeditation, or to intending to steal the car. We conclude that defendant voluntarily waived his Miranda rights and acknowledged killing the victims. Defendant contends that the deliberate police destruction of their notes of the Norwalk interview violated the requirements of People v. Hitch (1974) 12 Cal.3d 641 [117 Cal.Rptr. 9, 527 P.2d 361], requiring suppression of the Norwalk interview and arguably the Bakersfield interview as well.

We observe, however, that apart from admitting the killings, which were proven by eyewitness testimony in any case, defendant’s statements were largely exculpatory: the statements clearly did not admit all elements of first degree murder. Accordingly, they should be classed as admissions rather than confessions. The introduction of these statements was not prejudicial to defendant. 3.

Defendant points to numerous instances of prosecutorial misconduct, of which the most serious is the district attorney’s repeated references to Lufenberger’s testimony as “uncontradicted”; since defendant could not contradict that testimony without taking the stand, those references may constitute an impermissible comment upon defendant’s exercise of his right to avoid self-incrimination. Defense counsel, however, did not object to any of these comments. Since the harm could probably have been obviated by timely objection and admonition, we conclude that defendant is barred from raising this issue on appeal. 4.

In People v. Flannel (1979) 25 Cal.3d 668 [160 Cal.Rptr. 84, 603 P.2d 1], we held that in future cases a defendant is entitled (if the evidence warrants) to a sua sponte instruction that an honest but unreasonable belief in the necessity to act in self-defense may rebut malice and reduce a homicide to manslaughter. Since defendant Murtishaw’s trial antedated the finality of Flannel, the trial court did not err in failing to instruct sua sponte on the doctrine of unreasonable self-defense. 5.

The trial court, instructing the jury on the charge of assault with intent to commit murder, failed to explain to them that this crime requires express malice, and cannot rest upon theories of implied malice or felony murder. Its failure to so instruct was error (see People v. Mize (1889) 80 Cal. 41 [22 P. 80]; People v. Martinez (1980) 105 Cal.App.3d 938 [165 Cal.Rptr. 11]), but since overwhelming evidence demonstrates that defendant acted with express malice, the error is not prejudicial. 6.

The trial court denied a defense motion for discovery of prosecution investigations and records relating to the jury panel, or alternatively for funds to hire an investigator to investigate the prospective jurors. Although precedent supports the trial court’s order, the manifest unfairness of permitting a secret prosecution investigation of prospective jurors when the defense lacks funds to conduct a similar investigation demonstrates the need to reconsider our position on this issue.

In future cases the trial judge should have the discretion to grant the defense access to prosecution jury records and investigations. The value of such an investigation in the present case, however, is so speculative that the trial court’s ruling should not form a basis for reversible error. 7. At the penalty phase the prosecution called Dr. Ronald Siegel, a psychopharmacologist, who testified that defendant would probably engage in violent and perhaps homicidal behavior in a prison setting. As we explain, psychological forecasts of future violent behavior are of questionable reliability.

Moreover, such evidence, while only marginally relevant to the penalty jury’s task under former Penal Code

section 190.3, is likely to prove highly prejudicial. We therefore conclude that the trial court should not have permitted Dr. Siegel to testify as to his opinions concerning defendant’s future conduct in prison. The judgment imposing the death penalty must be reversed because the trial court erred in admitting over timely objection the penalty phase testimony of Dr. Siegel. The judgment of guilt is not affected by prejudicial error and is affirmed. I. Statement of facts. On the morning of April 9, 1978, defendant (age 20) and Gregory Lufenberger, defendant’s brother-in-law, decided to go shooting in the desert.

After acquiring two .22 rifles and purchasing cartridges and beer, they drove to the Town of Mojave. Leaving Mojave, after buying more beer, they drove down a dirt road in the desert. Defendant stalled the car and, apparently because of a defective starter, was unable to restart it. They remained by the car; defendant set up a beer can on the car, shot at it, and then fired several shots at the car itself.

In the early afternoon defendant and Lufenberger encountered four college students—Lance Bufflo, Martha Soto (Bufflo’s wife), James Henderson, and Ingrid Etayo—who had parked their car, a white Chevrolet, about two-tenths of a mile from defendant’s vehicle. The students were filming a movie for Bufflo’s cinema class at the University of Southern California. The plot involved a man (played by Henderson) stranded in the desert by car failure. He grows progressively weaker and is finally confronted by a hooded female figure (Etayo) symbolizing impending death.

He shoots at the figure, using a .38 revolver and blank cartridges, but is unable to harm it, and succumbs to his fate. Defendant and Lufenberger asked the students for a ride to a gas station to arrange car repairs; the students agreed to do so when they had finished filming. Defendant and Lufenberger then returned to their car and again attempted unsuccessfully to start it. During this period, according to Lufenberger, defendant “kept on saying, let’s shoot the people, or something like that.

He wanted to steal the car or something.” Defendant in his interview with the sheriffs office denied making any such statement. Later in the afternoon defendant and Lufenberger encountered the two women students driving by in the Chevrolet. They asked the women for a ride into town, but the women replied that they were not going to town, but only to pick up some lunch. Defendant and Lufenberger then returned to the site of the filming. Bufflo went over to talk to them, examined and test-fired defendant’s rifle, and showed them his rented .38 revolver, which he loaded with blanks.

Defendant watched him load the gun, but it is not clear whether he knew the cartridges were blanks. Defendant and Lufenberger again returned to their car; according to Lufenberger defendant again suggested shooting the students. Defendant and Lufenberger had been drinking beer steadily throughout the day, and had consumed from two to three 6-packs of beer. Bufflo and Lufenberger both described defendant as smelling of alcohol, but said that they observed no slurring of speech, unsteadiness, or other physical symptoms of extreme intoxication. About 5 p.m., the shadows grew too long for filming.

After completing the scene in which Henderson futilely shoots at the hooded figure, the students prepared to leave. Defendant followed the students back toward their car. As the students reached the car defendant, then about 80 feet away from the car, began to shoot at them. According to Bufflo, Henderson, who was carrying the revolver in a satchel, and Etayo reached the car first and went to the passenger side. As Bufflo and Soto walked to the driver’s side, Bufflo heard two or three shots. Henderson yelled out “I’ve been shot. Get me to a hospital.” Bufflo then heard five or six more shots strike the car.

Bufflo, Henderson (who had been wounded in the shoulder) and Etayo took refuge behind the passenger side of the car. Bufflo looked under the car and saw Soto lying on the ground. He and Henderson dragged her over to the passenger side. Soto had been shot in the head, an injury which caused her death two days later. After the pause in the firing during which the students attempted to help Soto, more shots struck the car. Etayo shouted that whoever was shooting should stop because people were wounded.

Henderson said they would all be killed unless someone went for help; he dashed out from behind the car and was struck and killed by five or six bullets. Bufflo crawled under the car to see who was shooting. He saw defendant standing behind a bush about 50 feet away. Defendant looked at him and fired several shots at him; one struck him in the hand, while the others struck the car. Bufflo told Etayo they would have to run. He ran along the road about 150 feet, tripped, and fell. Looking back, he saw defendant walking toward the car.

Bufflo ran further, turned, looked back, and saw defendant standing over Etayo, who was sitting next to Soto. Defendant yelled something at the women, but Bufflo could only make out the word “car.” He resumed running and heard more shots. Bufflo finally reached the highway, flagged down a van, and got a ride into Mojave where he notified the police. Lufenberger’s account of the killing differs in only minor respects from Bufflo’s testimony. Lufenberger said that he first heard a few shots, which sounded louder than shots from a .22. Lufenberger walked toward the Chevrolet.

He heard a number of shots from the .22, then saw defendant reload his gun and resume shooting. A hooded woman with an arm wound (presumably Etayo) came out from behind the car and walked toward defendant, begging him to stop shooting. Defendant told her to stop or he would shoot her. Lufenberger then ran back toward defendant’s car. He heard more shots. Defendant also ran back toward his car. When he caught up with Lufenberger, Lufenberger asked defendant why he had shot the students; defendant replied “I don’t know.” Lufenberger then asked if defendant intended to kill Lufenberger; defendant said no.

At Lufenberger’s suggestion they threw their guns in the desert and hitchhiked home. Defendant’s statement to the police presents a third version of the killing. He said he started toward the Chevrolet to get the promised ride into town. He then heard two or three shots. Knowing the students had finished the filming, he did not know what was happening; he felt confused, angry, and thought that the students were shooting at him. He immediately began firing at the car, emptying the entire clip.

As he reloaded, he heard Lufenberger shout “throw out your gun.” A shirtless man ran away (Bufflo); a hooded figure (Etayo) came at him from the car. He told the figure to stop and, when it did not, shot it. (Defendant did not mention shooting either Soto or Henderson.) After the shooting concluded, according to defendant, Lufenberger suggested that they take the Chevrolet, but defendant observed that fluid was leaking from the car where bullets had struck. He and Lufenberger then agreed to leave their guns in the desert and hitch a ride home. When the police arrived, Henderson and Etayo were dead.

Soto died two days later. All died of bullet wounds from a .22 rifle; ballistics tests indicated all shots came from defendant’s gun. Defendant and Lufenberger hitched a ride to a Denny’s Restaurant in Saugus, where they encountered their wives who had been looking for them. As they drove home to Norwalk together, defendant told his wife that their car had been stolen and that he had shot three people. They went to the home of defendant’s brother, Gerald Murtishaw, and told him of the killings.

With defendant’s permission, Gerald called the Norwalk police, who came and arrested defendant about 1 a.m. the morning of April 10. Kern County Sheriff’s Investigators Higgins and Pendleton arrived at the Norwalk police station about 6 a.m. After listening to a statement of his Miranda rights, defendant agreed to talk to the investigators; in response to questions he admitted killing the victims. The investigators did not record this interrogation, but Higgins took notes which he used to prepare a typed report, and then destroyed. The investigators then took defendant to Bakersfield.

Beginning about 2 p.m., the investigators, accompanied by a deputy district attorney, again interrogated defendant. The district attorney introduced the tape recording of the Bakersfield interview into evidence and supplied the jury with a transcript. Defendant and Lufenberger were charged with three counts of murder and one of assault with intent to commit murder. Both pled not guilty. Before preliminary hearing, the district attorney dismissed all charges against Lufenberger. Defendant was held to answer on all charges.

The superior court granted defendant’s request to enter a plea of not guilty by reason of insanity. It appointed three psychiatrists to examine defendant concerning his sanity at the time of the offense. It also granted the district attorney’s motion for appointment of psychopharmacologist Ronald Siegel to examine defendant to determine whether he was under the influence of PCP at the time of the crime. Shortly before trial, defendant filed a number of motions concerning the jury selection.

He first moved for discovery of any prosecution records or investigation of potential jurors, or in the alternative for funds to conduct a comparable defense investigation. He further requested the court either to forbid voir dire of the jurors concerning their views on capital punishment, or alternatively to empanel separate guilt and penalty juries, basing both requests on the theory that juries which exclude persons who would automatically vote against the death penalty are biased against defendant at the guilt phase of the trial (see Hovey v.

Superior Court (1980) 28 Cal.3d 1 [168 Cal.Rptr. 128, 616 P.2d 1301]). The trial court denied all motions, but ordered that voir dire of jurors concerning their views on the death penalty and their knowledge of pretrial publicity be conducted individually in chambers. At trial, defendant withdrew his plea of not guilty by reason of insanity and relied exclusively on a defense of diminished capacity. The prosecution therefore presented three expert witnesses on the question of defendant’s mental state at the time of the offense.

The two psychiatrists, Doctors Matychowiak and Burdick, stated that defendant was able to intend to kill, to premeditate, and to deliberate. They relied for their opinions upon his lack of history of mental illness, evidence that he was not severely inebriated, and his fairly accurate chronological recall of events. Both rejected the defense theory that he committed the crimes in a “fugue state,” a dissociative condition in which the actor sees himself as a sort of spectator to events he cannot control. Dr.

Siegel, a psychologist expert in drug-induced behavior, testified at length, explaining that in his opinion defendant’s conduct was not the result of his past use of PCP and other drugs. Dr. Siegel also opined that defendant was capable of premeditation at the time of the killing. Thus in essence the three prosecution experts painted a picture of defendant as a man under considerable psychological pressure at home (where he and his wife were living with her ex-husband), who has low tolerance for frustration, and who reacts to frustration by violence without caring about the consequences.

The only defense witness, Dr. Kelly, a psychiatrist, agreed in many respects with the prosecution experts’ analysis of defendant’s overall mental condition. Dr. Kelly, however, felt that defendant’s confused and dream-like description of the killing indicated that defendant at the time was in a “fugue state” in which he experienced a distorted sense of time and felt like he was outside his own body with no control over it. Dr. Kelly stated that while defendant was in this state, he would not be able to form any conscious intent and would be unable to premeditate or deliberate.

The jury found defendant guilty of the first degree murder of Henderson, Soto and Etayo, and of assault with intent to commit murder upon Bufflo. It further found the special circumstances of multiple murder. (Former Pen. Code, § 190.2.) At the penalty phase, the prosecution called Jerry Rothenberger, defendant’s former foreman.

Rothenberger testified over objection that about 10 days before the crime, he had told defendant to speed up his work; defendant became angry and said “just get out of here, just get out of here, I’m going to kill you, man, just go away.” Rothenberger did not testify that defendant accompanied his verbal threat with any physical act. The district attorney also recalled Dr. Siegel.

He testified over objection that defendant would continue to be a violent and assaultive person in prison, reacting to frustration with physical aggression, and could show the “same types of homicidal tendencies that he has shown in the past.” Following two days of deliberation, the jury returned a verdict of death. Defendant moved for modification of sentence.

The court reviewed what it felt were the aggravating factors—the multiple killings of vulnerable victims, the killing of Henderson as he fled, the killing of Etayo as she pled for mercy—and two mitigating factors—defendant’s youth and lack of a prior criminal record. Concluding that the aggravating factors outweighed the circumstances in mitigation, the court, making an independent determination that the verdict was supported by the evidence, denied the motion to modify the sentence. II. Substantial evidence supports the verdict finding defendant guilty of first degree murder.

The trial court instructed the jury that it could find defendant guilty of first degree murder on either of two theories: premeditated murder, or murder committed in the course of committing a felony; namely, an attempted robbery involving a theft of the victim’s car. We examine the evidence to support the verdict on both theories. We observe first that the evidence is ample to support a verdict finding premeditated murder. Three prosecution experts testified that defendant, although intoxicated to some degree, was capable of premeditation and deliberation at the time of the killings.

Although defendant argues that proof that he was able to premeditate does not prove that he did so, the testimonial account of the crime and events leading up to it strongly suggests that defendant did in fact premeditate the murders. In People v. Anderson (1968) 70 Cal.2d 15, 27 [73 Cal.Rptr. 550, 447 P.2d 942], we suggested three factors which might lead an appellate court to sustain a finding of premeditated murder: (1) facts showing prior planning activity; (2) facts suggesting motive; (3) facts about the manner of the killing which suggest a preconceived design.

The evidence in the present case, viewed in the light most favorable to the prosecution (see Jackson v. Virginia (1979) 443 U.S. 307, 319 [61 L.Ed.2d 560, 573, 99 S.Ct. 2781]) supports a finding of premeditation under the Anderson analysis. Lufenberger testified that earlier in the day the defendant suggested to Lufenberger that they kill the victims. Shortly before the killing he repeated that suggestion. Then, when the victims packed up their equipment and prepared to leave, defendant followed them to their car, carrying a loaded gun. About 80 feet from the car, defendant halted and began shooting.

After wounding Henderson and Soto (the latter fatally), he reloaded his gun, walked 30 feet closer, and resumed shooting. Defendant killed Henderson when Henderson attempted to flee. He then saw Bufflo looking at him from under the car, shot at Bufflo, and wounded him. Then when Etayo came out from behind the car and approached him, asking him to stop shooting, defendant according to his own statement warned her to stop. When she continued to approach he shot and killed her.

The foregoing chronicle shows both a preconceived plan to kill the victims, and a manner of killing which indicates careful and deliberate action. The fact that defendant paused at least once to reload his gun, and that he shot Etayo only after warning her not to approach him, strongly suggest that he was not in the grip of an impulsive, uncontrollable explosion of violence, but was able both before and during the killings to weigh and plan his actions. The only questionable factor in the Anderson analysis is that of motive.

Defendant may have intended to steal the car, as discussed subsequently in this section; he may have simply been venting his anger and frustration as the prosecution psychiatrists opined. But regardless of whether or not defendant acted with conscious motive, the evidence of planning and deliberate execution of his plan to kill the victims supports a finding of premeditated murder. The prosecution’s alternative theory of first degree murder—a killing in the course of a felony—presents a closer question. Indeed, the prosecution itself seemed to have doubts about this theory.

It did not charge defendant with attempted robbery or automobile theft, nor specify premeditated killing in the course of robbery as a special circumstance. It presented three expert witnesses who stated that in their opinion the defendant killed in reaction to cumulative frustrations; none said he did so in order to steal an automobile. Nevertheless, the court instructed the jury on a felony-murder theory, and the district attorney argued that theory to the jury.

The principal evidence to support that theory is Lufenberger’s testimony that defendant “kept on saying, ‘let’s shoot the people,’ or something like that. He wanted to steal the car or something.” When the victims prepared to leave, defendant, seeing his last opportunity to take the car and perhaps fearing that they would leave him and Lufenberger stranded in the desert, took his gun and followed them to the car. The public defender notes, on the other hand, that defendant had numerous earlier opportunities to steal the car, but did not attempt to do so.

Even when defendant began shooting at the victims, he did not demand possession of the car. By shooting at persons hiding behind and under the car, he risked disabling the vehicle. Finally, when the shooting was over he did not take the car, but his failure to do so may be explained by his statement that he saw something leaking from it. Weighing this evidence, we conclude that it is sufficient to support a verdict on a felony-murder theory. The jury was entitled to believe Lufenberger’s testimony which indicated that defendant was thinking of stealing the car.

It is true that the evidence cited by the public defender suggests that defendant would not have been satisfied with a nonviolent theft of the vehicle. It does not, however, necessarily indicate that defendant did not act with the concurrent goals of killing the victims and taking their car. III. Defendant’s statements to the sheriffs deputies were voluntary. Although a deputy’s destruction of the notes of the first interview arguably calls for sanctions under the doctrine of People v.

Hitch, supra, 12 Cal. 3d 641, since defendant’s statements were exculpatory the trial court’s failure to impose sanctions is not reversible error. About 6 a.m. the morning following the crime two investigators from the Kern County Sheriffs office questioned defendant at the Norwalk police station. Defendant affirmed that he understood his Miranda rights, and stated he was willing to talk to the investigators. During a 20- to 30-minute interview he explained his version of the events of the previous day. The investigators did not record this interview, but Investigator Higgins took handwritten notes.

Higgins later prepared a typed transcription, compared it with the handwritten notes for accuracy, and then destroyed the notes. Defendant was then taken to the Kern County Sheriff’s office in Bakersfield. At some point during the morning or afternoon, the investigators, joined by a deputy district attorney, resumed questioning defendant. Defendant again waived his Miranda rights and, in a two hour recorded interview, responded to questions concerning the crime.

Following the court’s denial of a motion to suppress, Deputy Higgins testified briefly to the Norwalk interview, and the prosecution introduced a tape recording of the Bakersfield interview. Defendant argues that both interviews were inadmissible on two closely related grounds: that defendant did not voluntarily waive his Miranda rights before responding to the deputies’ questions and that his responses themselves were involuntary.

He further contends that Deputy Higgins’ destruction of his notes of the Norwalk interview constitutes an unlawful suppression of evidence which requires exclusion of his statements to the deputies. We begin by considering the question whether defendant’s waiver of Miranda rights and ensuing statements to the deputies were involuntary. In People v. Jimenez (1978) 21 Cal.3d 595 [147 Cal.Rptr. 172, 580 P.2d 672], we held that the prosecution must prove the voluntariness of a confession beyond a reasonable doubt. (21 Cal. 3d at p. 608.) That requirement applies equally to admissions by a defendant (see People v.

Alfieri (1979) 95 Cal.App.3d 533, 542 [157 Cal.Rptr. 304]), and, since confessions or admissions obtained without a valid waiver of Miranda rights are deemed to have been coerced, the prosecution bears also the burden of proving waiver beyond a reasonable doubt. (People v. Daniels (1969) 1 Cal.App.3d 367, 373-374 [81 Cal.Rptr. 675].) On appeal, it is our “duty to examine the uncontradicted facts to determine independently whether the trial court’s conclusion of voluntariness was properly found.” (People v. Jimenez, supra, 21 Cal.3d 595, 609, quoting People v.

Sanchez (1969) 70 Cal.2d 562, 571-572 [75 Cal.Rptr. 642, 451 P.2d 74].) Defendant testified without contradiction that at the time of the Nor-walk interview he had been without food or sleep for approximately 24 hours. In fact, he claimed he had neither food nor sleep until after the taped interview in Bakersfield, a duration of about 32 hours. The investigating officers testified without contradiction that during both interviews defendant was alert, responsive, and did not appear intoxicated; they stated that defendant did not complain of lack of food or sleep or request anything from the officers.

On this record, we conclude that we should uphold the trial court’s finding of voluntariness. Two factors weigh heavily with us in reaching this conclusion. First, defendant showed no unwillingness to admit to the killings. Before he was arrested defendant had already done so to his wife, his brother, and Lufenberger’s wife. When interrogated by the investigators, he cooperated without hesitation. Second, in contrast with some of the cases cited by the public defender, defendant’s lack of food or sleep was not the result of police misconduct.

In fact, until they questioned defendant, the investigators had no way to know when defendant last ate or slept. Under these circumstances, defendant had no reason to believe that he had to talk to the deputies before he could get food or an opportunity to sleep. The deputies’ testimony that defendant was alert and responsive during the interviews—testimony not contradicted by defendant—suggests that the lack of food or sleep, did not leave defendant unable to understand his constitutional rights, to voluntarily waive those rights, or to voluntarily respond to questions from the deputies.

Defendant’s claim that evidence was improperly suppressed presents a different issue. In People v. Hitch (1974) 12 Cal.3d 641 [117 Cal.Rptr. 9, 527 P.2d 361], defendant moved to suppress a breathalyzer test result on the ground that the police destruction of the test ampoule deprived him of due process.

We stated that if there was a reasonable possibility that the evidence was favorable to the defense, due process requires its disclosure. (Pp. 649-650.) When “such evidence cannot be disclosed because of its intentional but nonmalicious destruction by the investigative officials, sanctions shall in the future be imposed ... unless the prosecution can show that the governmental agencies involved have established, enforced and attempted in good faith to adhere to rigorous and systematic procedures designed to preserve the [evidence].” (Pp. 652-653.) The appropriate sanction, we ruled, was not a dismissal of the action but exclusion of the breathalyzer results. (P. 653.) The Attorney General argues that Hitch is inapplicable to the present case because the investigating officers preserved the results of the Norwalk interrogation through Higgins’ transcribed notes.

The investigator’s failure to record the interview and their destruction of the handwritten notes, however, made it impossible for the defense to verify whether the typed transcript reflects the handwritten notes or the reality of the interrogation. In this respect the present case closely resembles Hitch, in which the police preserved the results of the breathalyzer test, but effectively precluded the defendant from verifying the accuracy of those results. Two Court of Appeal opinions discuss analogous issues but point to different results. In People v.

Goss, supra, 109 Cal.App.3d 443, the police accidently destroyed a tape recording of a confession; the Court of Appeal found a violation of Hitch and upheld suppression of the confession. The court in In re Gary G. (1981) 115 Cal.App.3d 629 [171 Cal.Rptr. 531], on the other hand, ruled that police destruction of rough notes recording an interview with a witness did not violate Hitch. We need not resolve this implicit conflict because in the present context the admission of defendant’s statements, whether erroneous or not, could not constitute reversible error.

Those statements, to the extent that they differed from the testimony of Lufenberger and Bufflo, were primarily exculpatory. Defendant stated that he shot in response to gunfire which, he thought, was aimed at him. He denied any intent to kill the victims or to steal their car. He described generally a confused state of mind and impulsive action, a description which implies a failure to premeditate and suggests lack of malice aforethought. Thus with respect to the charge of first degree murder, the statements of defendant constitute admissions rather than confessions. In People v.

McClary (1977) 20 Cal.3d 218, 230 [142 Cal.Rptr. 163, 571 P.2d 620], we described “a confession as amounting to a declaration of defendant’s intentional participation in a criminal act, whereas an admission is merely a recital of facts tending to establish guilt when considered with the remaining evidence in the case.” A confession must encompass all elements of the crime. (People v. Fuentes (1967) 253 Cal.App.2d 969, 975 [61 Cal.Rptr. 768]; People v.

Beverly (1965) 233 Cal.App.2d 702, 712-713 [43 Cal.Rptr. 473].) Thus, although defendant’s statements might conceivably be considered a confession to manslaughter, they cannot be viewed as a confession to first degree murder. (Cf. People v. Fitzgerald (1961) 56 Cal.2d 855 [17 Cal.Rptr. 129, 366 P.2d 481], in which defendant’s statements constituted a confession to robbery, but an admission with respect to the charge of murder.) Although the improper introduction of a confession is reversible error per-se (People v.

Randall (1970) 1 Cal.3d 948, 958 [83 Cal.Rptr. 658, 464 P.2d 114]), the wrongful introduction of an admission is not reversible if the People can show beyond a reasonable doubt that the error complained of did not contribute to the verdict. (People v. McClary, supra, 20 Cal.3d 218, 230.) We believe that the People have met this burden. Uncontradicted eyewitness testimony of Bufflo and Lufenberger established that defendant fired the fatal shots. As we explained, defendant’s statements, to the extent that they contradict the eyewitness testimony, are exculpatory.

The exclusion of those statements would merely tip the scale more heavily in favor of the prosecution theory that defendant committed premeditated murder for the purpose of stealing the victim’s car, and would reduce the evidence of defendant’s confused state of mind and intoxication, evidence which supported defendant’s claim of diminished capacity. We therefore conclude that even if the admission of defendant’s statements was error, such error was harmless. IV. Defendant waived his right to object to prosecutorial misconduct.

The public defender recites a number of instances of alleged misconduct by the prosecutor, and contends that the cumulative effect of this misconduct prejudiced the defense below. The most serious of these allegations concern comments by the prosecutor in his guilt phase argument to the jury which assertedly reflect on defendant’s failure to testify. The public defender refers to five such statements. The first, quoted in the footnote below, probably referred not to defendant’s failure to testify, but to omissions in defendant’s statement to the police.

The remaining comments, however, all relate to defendant’s failure to contradict Lufenberger’s testimony that defendant “kept on saying, let’s shoot the people, or something like that. He wanted to steal the car or something.” Prosecutorial comment which draws attention to a defendant’s exercise of his constitutional right not to testify, and which implies that the jury should draw inferences against defendant because of his failure to testify, violates defendant’s constitutional rights. (Griffin v.

California (1965) 380 U.S. 609 [14 L.Ed.2d 106, 85 S.Ct. 1229].) Applying that principle, federal courts have “held that for the government to say, in summation to the jury, that certain evidence was ‘uncontradicted,’ when contradiction would have required the defendant to take the stand, drew attention to his failure to do so, and hence was unconstitutional.” (United States v. Flannery (1st Cir. 1971) 451 F.2d 880, 881; see United States v. Buege (7th Cir. 1978) 578 F.2d 187, 188.) California decisions reach the same result. In People v.

Vargas (1973) 9 Cal.3d 470 [108 Cal.Rptr. 15, 509 P.2d 959], the prosecutor commented that “there is no denial at all that they [defendants] were there”; we held that comment improperly reflected on defendants’ failure to testify. (See 9 Cal.3d at pp. 476-477.) In People v. Medina (1974) 41 Cal.App.3d 438 [116 Cal.Rptr. 133], the prosecutor said the testimony of his witnesses was “unrefuted”; the Court of Appeal found Griffin error because “the defendants, who were the only ones who could have refuted it, did not take the stand.” (41 Cal.App.3d at p. 457; see People v.

Northern (1967) 256 Cal.App.2d 28, 30-31 [64 Cal.Rptr. 15].) Defense counsel failed to object to any of the asserted improper comments of the district attorney. Recently in People v. Green, supra, 27 Cal.3d 1, we had occasion to reconsider the standard to be determined in deciding whether failure to object to proscutorial misconduct bars defendant from raising that misconduct as error on appeal.

Overruling prior cases, we explained that “the initial question to be decided in all cases in which a defendant complains of prosecutorial misconduct for the first time on appeal is whether a timely objection and admonition would have cured the harm.

If it would, the contention must be rejected...; if it would not, the court must then and only then reach the issue whether on the whole record the harm resulted in a miscarriage of justice ....”(P. 34.) In our opinion, a timely objection and admonition, when the district attorney first referred to Lufenberger’s testimony as “uncontradicted,” would probably have cured the effect of the misconduct. We acknowledge that repeated objections and admonitions might well, as the public defender asserts, merely emphasize to the jurors the defendant’s failure to testify.

We have no reason, however, to assume that the district attorney would have continued to make improper references to the “uncontradicted” testimony once the court has sustained an objection to such a comment. V. People v. Flannel, supra, 25 Cal.3d 668, does not require a sua sponte instruction on the doctrine of unreasonable self-defense in cases tried prior to the finality of that opinion. In People v.

Flannel (1979) 25 Cal.3d 668 [160 Cal.Rptr. 84, 603 P.2d 1], we reaffirmed the doctrine of unreasonable self-defense, holding that in cases tried subsequent to the finality of Flannel the court must instruct sua sponte, when the evidence warrants, that “one who holds an honest but unreasonable belief in the necessity to defend against imminent peril to life or great bodily injury does not harbor malice and commits no greater offense than manslaughter.” (P. 672.) Defendant contends that, even though the present case was tried before Flannel, the trial court’s failure to instruct sua sponte on this doctrine is reversible error.

If the present case had been tried subsequent to Flannel, it is clear that the court would be required to give a sua sponte instruction. Defendant told the investigators that when he approached the victims “something went bang and it came toward me and ... I just started shooting back in that direction.” Defendant claimed to have heard Lufenberger “yelling at the people saying ‘throw out your gun.’ .... [Ajbout that time someone came running from the car towards me. I didn’t know exactly, you know, if they had a gun or what.. .. ” Defendant then shot the person who was approaching him.

This evidence suggests that when defendant shot the victims, he believed he was acting in self-defense, returning the fire of persons who had been shooting at him. The trial court therefore instructed the jury on reasonable self-defense as a complete defense to homicide. The jury, however, would be more likely to find that defendant’s fear of assault was unreasonable, since defendant by his own statement never saw any victim aim a gun in his direction.

Even if shots were fired as defendant approached the victim’s car—a point as to which the evidence is in conflict—the jury could conclude that it was unreasonable for defendant to believe they were aimed at him. The record therefore contains substantial evidence to support a finding that defendant acted under an unreasonable belief that his life was in danger, a finding which under Flannel would reduce the crime from murder to manslaughter. The court would therefore be required to instruct the jury sua sponte to that effect. (See People v.

Flannel, supra, 25 Cal.3d 668, 682, 683, fn. 9.) Our Flannel decision, however, stated that the requirement for a sua sponte instruction did not apply to cases, such as the present case, tried before Flannel became final. We reasoned that the trial court need instruct sua sponte only on established general principles of law, for it would impose too onerous a burden to require the court to research and expound obscure doctrines not presented by the litigants.

In view of the hitherto undeveloped state of the unreasonable self-defense doctrine, we concluded that failure to instruct on it sua sponte was not error in cases not yet tried. (25 Cal.3d at pp. 681-683.) Defendant argues that we erred in limiting the retroactive effect of our Flannel decision since that decision did not announce new law, but reaffirmed a doctrine implicit in prior cases. This argument mistakes the basis of our discussion in Flannel.

We did not hold that the doctrine of unreasonable self-defense was a new doctrine which applies only prospectively; to the contrary, we made clear that even in prior cases it would be error for the court to refuse a request to instruct on unreasonable self-defense. We held, instead, that the doctrine was so obscure and undeveloped that the trial court did not err if it failed to discover and instruct sua sponte on that principle.

We therefore reaffirm our conclusion in Flannel that, in cases tried prior to the finality of that decision, the trial court had no duty to instruct sua sponte oh unreasonable self-defense. The State Public Defender argues we should follow our decision in People v. Wheeler (1978) 22 Cal.3d 258 [148 Cal.Rptr. 890, 583 P.2d 748], when in holding that a prosecutor could not use peremptory challenges to exclude all members of a race from the jury, we stated that “[t]he rule we adopt herein applies to the defendants in the case at bar and in the companion matter of People v.

Johnson . .. and to any defendant now or hereafter under sentence of death. In all other cases the rule will be limited to voir dire proceedings conducted after the present decision becomes final.” (P. 283, fn. 31.) (Italics added.) On authority of Wheeler, he points out, we reversed the conviction in People v. Allen (1979) 23 Cal.3d 286 [152 Cal.Rptr. 454, 590 P.2d 30], even though that case was tried before Wheeler became final. The State Public Defender, however, once again overlooks the specific holding in Flannel.

He argues as if we held that the trial judge in Flannel erred by failing to instruct sua sponte on unreasonable self-defense, but refused to give that holding retroactive effect. In fact, we held that the trial judge did not err, and that, until the doctrine of unreasonable self-defense received elucidation by the filing of our opinion, trial courts were not under a duty to instruct on that doctrine sua sponte. Applying this reasoning, we further held that Flannel himself could not predicate error on the court’s failure to instruct sua sponte at his trial.

Thus even if we applied Flannel retroactively to the present case, defendant here would be entitled to no greater benefit than that accorded Flannel himself; he would not be entitled to posit error on the trial court’s failure to instruct sua sponte. Finally, the State Public Defender argues that under Beck v. Alabama (1980) 447 U.S. 625 [65 L.Ed.2d 392, 100 S.Ct. 2382] we are constitutionally required to give retroactive effect to Flannel's requirement for a sua sponte instruction. This contention far overstates the reach of the high court decision.

Beck concerned an Alabama statute which required a jury in a capital case to choose between the death penalty and acquittal, without considering lesser included offenses.

The Supreme Court held the statute unconstitutional, on the ground that “failure to give the jury the ‘third option’ of convicting on a lesser included offense would seem inevitably to enhance the risk of an unwarranted conviction.” (447 U.S. at p. 637 [65 L.Ed.2d at pp. 402-403, 100 S.Ct. at p. 2389].) In contrast to Beck, the court in the present case instructed the jury on the elements of the pertinent lesser included offenses and made it clear that the jury could return a verdict finding defendant guilty of such offenses.

The risk of an unwarranted conviction forced by an all- or-nothing choice between death and acquittal—the danger presented in Beck—is absent in this case. VI. Contradictory jury instructions defining the crime of assault with intent to commit murder do not give rise to prejudicial error. Defendant was convicted of assault with intent to commit murder (Pen. Code, § 217) on Lance Buffio, the surviving victim. The jury instructions relating to this crime paint a confusing picture.

First, the jury was instructed in terms of CALJIC No. 3.31 that each of the crimes charged (murder and assault with intent to commit murder) require a specific intent, and that “[i]n the crime of first degree murder, in the crime of second degree murder based on express malice aforethought and in the crime of assault with intent to commit murder, the necessary specific intent is to unlawfully kill.” From this instruction the jury would have to conclude that it could not convict defendant on the assault charge without finding that he intended to kill the victims.

The jury was also instructed under the 1970 revision of CALJIC No. 9.01 that “[ejvery person who assaults another with the specific intent to commit murder” is guilty of a violation of Penal Code

section 217. This instruction speaks of intent to “murder,” not to “kill,” and as, we shall explain, the instructions defined murder to include forms of murder not requiring an intent to kill. Specifically, the court instructed the jury it could find defendant guilty of murder on any of three theories: express malice, implied malice, or felony murder. Express malice, it explained, means “an intention unlawfully to kill.” Malice is implied, it instructed, “when the killing results from

an act involving a high degree of probability that it will result in death, which act is done for a base, antisocial purpose and with a wanton disregard for human life by which is meant an awareness of a duty imposed by law not to commit such acts followed by the commission of the forbidden act despite that awareness or when the killing is a direct causal result of the perpetration or the attempt to perpetrate a felony inherently dangerous to human life.” (CALJIC No. 8.11.) Thus, taken as a whole, the instructions on assault with intent to commit murder were contradictory.

The court defined the mental element essential to the crime in two different ways—intent to kill and intent to murder—and by implication defined the latter to include forms of murder not requiring an . intent to kill. Defendant contends that the trial court erred when it failed to instruct that only express malice could support a conviction for assault with intent to commit murder. The court should have limited its definition of implied malice and felony murder, defendant claims, by explaining that those concepts are inapplicable to the assault charge. We agree with defendant’s contention.

The concept of felony murder is clearly inapplicable to the crime of assault with intent to commit murder (People v. Heffington (1973) 32 Cal.App.3d 1 [107 Cal.Rptr. 859]); the Attorney General concedes as much. Established California authority further demonstrates that the concept of implied malice, insofar as it permits a conviction without proof of intent to kill, is also inapplicable to a charge of assault with intent to commit murder. In People v.

Mize (1889) 80 Cal. 41 [22 P. 80], the court instructed the jury that defendant would be guilty of assault with intent to commit murder if his acts were such that he could have been convicted of murder had the victim died.

The court held the instruction erroneous: ‘“To constitute murder, the guilty person need not intend to take life; but to constitute an attempt to murder, he must so intend.’ [Citation.] ‘The wrongdoer must specifically contemplate taking life; and though his act is such as, were it successful, would be murder, if in truth he does not mean to kill, he does not become guilty of an attempt to commit murder.’” (80 Cal. at p. 43.) (Accord, People v. Miller (1935) 2 Cal.2d 527, 532-533 [42 P.2d 308].) The recent decision of the Court of Appeal in People v.

Martinez (1980) 105 Cal.App.3d 938 [165 Cal.Rptr. 11], continues the line of authority dating to People v. Mize, supra. In Martinez the trial court, in a case involving assault with intent to commit murder, instructed on implied malice and felony murder; Martinez held those instructions constituted reversible error. (105 Cal.App.3d at p. 943.) The Attorney General asserts that Martinez was decided erroneously.

He argues that although the crime of assault with intent to commit murder requires both an intent to kill and malice aforethought (to make the killing a murder), that malice can be either express or implied. Thus he concludes that it is not erroneous to instruct on implied malice in a case involving assault with intent to commit murder. The problem with the Attorney General’s argument is that once a defendant intends to kill, any malice he may harbor is necessarily express malice. Implied malice, as defined in CALJIC No. 8.11, cannot coexist with a specific intent to kill.

To instruct on implied malice in that setting, therefore, may confuse the jury by suggesting that they can convict without finding a specific intent to kill. The court’s erroneous instructions in the present case, however, were not prejudicial. Defendant was convicted of the first degree murders of Martha Soto, James Henderson, and Ingrid Etayo, and in light of the evidence presented it is virtually certain that the jury found defendant intended to kill these victims. Defendant’s assault on Lance Bufflo cannot be distinguished from his slaying of Bufflo’s companions.

Furthermore, Bufflo specifically testified that when defendant saw Bufflo looking at him from under the Chevrolet, he aimed and fired directly at Bufflo. It is inconceivable that the jury, having rejected defendant’s claim of diminished capacity, would find that defendant had no intent to kill. Applying the test of prejudice established in People v. Watson (1956) 46 Cal.2d 818, 836 [299 P.2d 243]—whether it is reasonably probable that a result more favorable to defendant would have been reached in the absence of the error—we find the error nonprejudicial. VII.

In future cases a trial court should have the discretion to permit a defendant, who lacks funds to investigate prospective jurors, to inspect prosecution jury records and investigations. On December 12, 1978, several weeks before trial, defense counsel moved in the alternative for discovery of prosecutorial investigations of prospective jurors or for $1,000 to enable the defense to conduct a similar investigation. Defendant renewed the motion on December 29, just before the start of jury selection.

The district attorney at that time acknowledged that his office had conducted field investigations of prospective jurors and maintained records showing how the jurors had voted in prior cases and whether they had arrest records. The trial court, however, denied defendant’s motion. As defendant recognizes, all California decisions and a majority of decisions from other jurisdictions have held that a trial court did not err in denying a defendant access to prosecution jury records. (See People v. Brawley (1969) 1 Cal.3d 277, 293-294[82 [82 Cal.Rptr. 161, 461 P.2d 361] and cases there cited; People v.

Castro (1979) 99 Cal.App.3d 191, 198-199 [160 Cal.Rptr. 156]; People v. Airheart (1968) 262 Cal.App.2d 673, 679-681 [68 Cal.Rptr. 857]; People v. Darmiento (1966) 243 Cal.App.2d 358 [52 Cal.Rptr. 428]; People v. Superior Court (1959) 175 Cal.App.2d 830 [1 Cal.Rptr. 55, 78 A.L.R.2d 306]; People v.

Ruef (1910) 14 Cal.App. 576, 596 [114 P. 48, 54]; see generally Annot. (1978) 86 A.L.R.3d 571.) The cases rest on two propositions: (1) it is uncertain whether the defendant would derive any significant advantage from discovery of jury investigations and records; (2) much of the information the defendant seeks could be learned by voir dire of the individual jurors. Neither proposition supports the denial of discovery.

It may be doubtful whether public funds should be spent on investigations which yield no clear and certain benefit, especially if the same information could be discovered in voir dire, but those doubts cannot justify making the results of the investigation available to one side but not to the other. Whatever doubts the courts may have, it is apparent that the prosecutor here believes the advantage he gains from jury investigations and records justifies the expense.

When courts then deny defendants who cannot afford similar investigations access to the prosecutor’s records, the result is that prosecutors in case after case will have substantially more information concerning prospective jurors than do defense counsel. Such a pattern of inequality reflects on the fairness of the criminal process. We therefore hold, under our authority to supervise the administration of California criminal procedure (see Hovey v.

Superior Court (1980) 28 Cal.3d 1, 80 [168 Cal.Rptr. 128, 616 P.2d 1301] and cases there cited), that following the finality of this opinion a trial judge will have discretionary authority to permit defense access to jury records and reports of investigations available to the prosecution. The foregoing holding does not require us to reverse the conviction in the present case. As the prior cases have pointed out, in any individual case it is entirely speculative whether denial of access caused any significant harm to the defense.

Consequently, under the test of prejudice established in the California Constitution (art. VI, § 13) and People v. Watson, supra, 46 Cal.2d 818, 836, the denial of access is not reversible error. VIII. The trial court erred in admitting the penalty phase testimony forecasting that defendant would commit violent and possibly homicidal acts in prison. The prosecution recalled Dr. Ronald Siegel, a psychopharmacologist, to testify in the penalty trial.

He stated over objection that defendant in a prison setting “will continue to be a violent assaultive and combative individual, and in constraints, with restraints on his behavior, and in a custodial care situation such as in a prison or in a jail, because of his inability to relate inter-personally, because of his latent rage and hostility and violence, I believe that this will be manifested in repeated displays.

I think these displays, in my opinion, will be physical rather than verbal simply because his history, his past history and my clinical studies of him have indicated, his past history is one of physical rather than verbal displays. And that he may become not only assaultive and violent, but he could show the same types of homicidal tendencies that he has shown in the past, with no ability to morally or physically constrain himself to the demands of the environment in which he finds himself.” We believe that the trial court should not have permitted this testimony.

We shall explain that (1) expert predictions that persons will commit future acts of violence are unreliable, and frequently erroneous; (2) forecasts of future violence have little relevance to any of the factors which the jury must consider in determining whether to impose the death penalty; (3) such forecasts, despite their unreliability and doubtful relevance, may be extremely prejudicial to the defendant. The admission of this testimony, in our opinion, constitutes error requiring reversal of the verdict at the penalty trial.

Numerous studies have demonstrated the inaccuracy of attempts to forecast future violent behavior. Two commentators summarized the results as follows; “Whatever may be said for the reliability and validity of psychiatric judgments in general, there is literally no evidence that psychiatrists reliably and accurately can predict dangerous behavior.

To the contrary, such predictions are wrong more often than they are right.” (Ennis & Litwack, Psychiatry and the Presumption of Expertise: Flipping Coins in the Courtroom (1974) 62 Cal.L.Rev. 693, 737.) Professor Dershowitz in 1969 pointed to the skewed results characteristic of psychiatric forecasts: “it seems that psychiatrists are particularly prone to one type of error—over-prediction .... [F]or every correct psychiatric prediction of violence, there are numerous erroneous predictions.” (Dershowitz, The Psychiatrist's Power in Civil Commitment: A Knife That Cuts Both Ways (Feb. 1969) Psych.

Today, at p. 47.) Cocozza and Steadman in 1976 reviewed the various studies and reported that “Whether one examines the legal, behavioral science or psychiatric literature on predictions of dangerousness, one constantly encounters conclusions similar to the one reached by Dershowitz that psychiatrists are generally inaccurate predictors.” (Cocozza & Steadman, op. cit., supra, 29 Rutgers L.Rev. at p. 1085.) In 1978 Professor Monahan undertook a further review of studies of violence prediction and noted that the percentage of false positives (erroneous predictions that a subject would engage in violent behavior) never fell below 54 percentage and went as high as 99.7 percent. ' (Monahan, The Prediction and Control of Violent Behavior (1978) pp. 179-196, in Hearings Before the House Subcom. on Domestic and International Scientific Planning, 95th Cong. 2d Sess., pp. 175-252.) We took judicial notice of studies showing the unreliability of attempts to forecast violence in People v.

Burnick, supra, 14 Cal.3d 306. We there observed that: “During the past several years further empirical studies have transformed the earlier trend of opinion into an impressive unanimity: ‘The evidence, as well as the consensus of opinion by responsible scientific authorities, is not unequivocal.’ (Diamond, The Psychiatric Prediction of Dangerousness (1975) 123 U.Pa. L.Rev. 439, 451.) In the words of spokesmen for the psychiatric profession itself, ‘Unfortunately, this is the state of the art.

Neither psychiatrists nor anyone else have reliably demonstrated an ability to predict future violence or “dangerousness.” Neither has any special psychiatric “expertise” in this area been established.’ (Task Force Report, Clinical Aspects of the Violent Individual (American Psychiatric Assn., 1974) p. 28.) And the same studies which proved the inaccuracy of psychiatric predictions have demonstrated beyond dispute the no less disturbing manner in which such prophecies consistently err: they predict acts of violence which will not in fact take place (‘false positives’), thus branding as ‘dangerous’ many persons who are in reality totally harmless. (See generally id. at pp. 23-30.)” (P. 327.) In Burnick, we cited the uncertain character of psychological prediction to support our conclusion that a person should not be committed as a mentally disordered sex offender unless found to be such beyond a reasonable doubt.

The issue in Tarasoff v. Regents of University of California (1976) 17 Cal.3d 425 [131 Cal.Rptr. 14, 551 P.2d 334, 83 A.L.R.3d 1166], was different—whether a psychotherapist who predicted that a patient would kill an identifiable victim owed a duty to exercise reasonable care to avert that danger. Rejecting the contention that a therapist should not act on the basis of predictions of such demonstrated unreliability, we held that when a therapist predicts or should predict a danger of violence, he incurs an obligation to use reasonable care to protect the intended victim.

We explained in Tarasoff that imposition of such a duty of care was not inconsistent with our decision in Burnick. In contrast to Burnick the issue in Tarasoff, we said, was “not whether the patient should be incarcerated, but whether the therapist should take any steps at all to protect the threatened victim; some of the alternatives open to the therapist, such as warning the victim, will not result in the drastic consequences of depriving the patient of his liberty.

Weighing the uncertain and conjectural character of the alleged damage done the patient by such a warning against the peril to the victim’s life, we conclude that professional inaccuracy in predicting violence cannot negate the therapist’s duty to protect the threatened victim.” (17 Cal.3d at p. 439.) Thus in Tarasoff we balanced the “uncertain and conjectural” harm to the patient against the mortal risk to the potential victim, and concluded that the therapist should act on the basis of his prediction, unreliable though it may be. That same balancing process in the present context yields a far different result.

There is nothing speculative about the harm to defendant, who faces not merely a risk of short-term incarceration but of execution. What is uncertain and conjectural is whether defendant, if imprisoned for life, will at some uncertain future date assault some yet unidentified victim. The calculus of risk which called for acting despite uncertainty in the Tarasoff setting does not justify executing a defendant to avoid improbable and speculative danger.

The Attorney General points out that, despite our doubts concerning psychological predictions of future violence, we have never barred the expert witness from testifying to his opinion on that subject. It would be consistent with past cases, he suggests, to hold that difficulties in prediction go to the weight of the evidence rather than to its admissibility. We reject this suggestion, for two reasons. First, as we recognized earlier, “the penalty of death is qualitatively different from a sentence of imprisonment, however, long....

Because of that qualitative difference, there is a corresponding difference in the need for reliability in the determination that death is the appropriate punishment in a specific case.” (Woodson v. North Carolina (1976) 428 U.S. 280, 305 [49 L.Ed.2d 944, 961, 96 S.Ct. 2978] (plur. opn. of JJ.

Stewart, Powell, and Stevens.) Consequently, “[although the utilization of predictive testimony by mental health experts in death penalty hearings does not differ in many ways from reliance upon it in other contexts, the unique severity of the penalty increases the importance of avoiding unjustified reliance upon predictions of ‘dangerousness.’” (Dix, The Death Penalty, “Dangerousness,” Psychiatric Testimony, and Professional Ethics (1977) 5 Am. J. Crim.

L. 151, 212.) In short, evidence which is barely reliable enough to justify a civil judgment or a limited commitment is not reliable enough to utilize in determining whether a man should be executed. Second, in most of the other cases in which courts have upheld admission of opinion testimony forecasting future violence, such as People v. Henderson (1980) 107 Cal.App.3d 475, 484 [166 Cal.Rptr. 20] the trier of fact is required by statute to determine whether a person is “dangerous.” (See, e.g., Pen. Code, § 1026.2 (restoration to sanity); Welf. & Inst.

Code, §§ 5276, 5300 (Lanterman-Petris-Short Act); Welf. & Inst. Code, § 6300 (mentally disordered sex offender hearings).) In such cases expert prediction, unreliable though it may be, is often the only evidence available to assist the trier of fact. The penalty jury in a murder trial, however, is not required to determine whether the defendant is dangerous or likely to commit future violence; in fact, as we shall explain, such a determination is at best only marginally relevant to the task at hand. Expert forecasts of future violent acts, as we have said, have little relevance to the determination of penalty.

The Legislature has listed specifically the factors which the jury must consider in deciding whether to impose the death penalty. The testimony of Dr. Siegel is not relevant to any of the listed factors. Former

section 190.3, however, permitted the parties to introduce evidence “as to any matter relevant to aggravation, mitigation, and sentence, including, but not limited to, the nature and circumstances of the present offense, the presence or absence of other criminal activity by the defendant which involved the use or attempted use of force or violence or which involved the expressed or implied threat to use force or violence, and the defendant’s character, background, history, mental condition and physical condition.” The admission of evidence is not limited to matters relevant to the specified aggravating or mitigating factors.

Arguably the risk that defendant will commit future violent acts is a matter “relevant to aggravation” which reflects on defendant’s future “character” and “mental condition.” In view of the unreliability of that forecast, however, the probative value of that testimony is far outweighed by its prejudicial impact. One can imagine few matters more prejudicial at the penalty trial than testimony from an established and credentialed expert that defendant, if sentenced to life without possibility of parole, would be likely to kill again.

Such testimony implants in the mind of each juror the message that the death penalty, promptly carried into effect, is the only way to protect society from the defendant—the only way to forestall another instance in which defendant responds to frustration with deadly violence. “A trier of fact offered the opportunity to base a decision on the affirmative assertion by an apparently well-qualified professional that a defendant’s execution is essential to saving the lives of others is likely to take that opportunity rather than face the difficult task of evaluating the offender’s ethical culpability. ... [Such predictions tend] to discourage the sort of individualization that the Supreme Court has demanded of death penalty procedures.” (Dix, Participation by Mental Health Professionals in Capital Murders Sentencing (1978) 1 Internat.

J. of L. & Psych. 283, 397.) We acknowledge that despite the recognized general unreliability of psychiatric predictions concerning future violence, it may be possible for a party in a particular case to show that a reliable prediction is possible. A more reliable forecast, for example, might be possible if the psychiatrist had established a close, long-term relationship with defendant that gives him a greater understanding of defendant’s behavior thpn can usually be attained in brief, often adversary, pretrial interviews.

A reliable prediction might also be conceivable if the defendant had exhibited a long-continued pattern of criminal violence such that any knowledgeable psychiatrist would anticipate future violence. Thus, we do not adopt an absolute rule barring such predictions at the penalty phase of a capital trial. In the present case, however, Siegel examined defendant on only one occasion—an examination devoted largely to determining whether defendant acted under the influence of PCP or other drugs—and had no established and close relationship with defendant on which to base his prediction.

Defendant’s asserted past violent acts were few and relatively trivial. Under these circumstances we have no reason to believe that Siegel’s prediction was immune from the general unreliability which attends predictions of future violence generally. In sum, Siegel’s prediction that defendant would be violent in a prison setting is unreliable. In fact, judging by the studies cited, that prediction is probably wrong; in other words, despite Siegel’s forecast, it is statistically more probable that defendant will not commit such violent acts.

The prediction is not directly relevant to any of the factors which the jury was required to consider in fixing the penalty under former

section 190.3, and, although arguably admissible under that statute, the prejudicial impact of that testimony far outweighs its probative value. We conclude that the trial court should not have permitted Siegel to testify over objection as to his opinion respecting defendant’s future behavior. The erroneous admission of Siegel’s testimony, is prejudicial error under any standard of review. Siegel’s prediction that defendant would be violent and homicidal in prison was the principal prosecution penalty phase evidence.

The jury deliberated two full days before deciding on the death penalty, suggesting that the issue of penalty was close. Under these circumstances, even if we applied the most liberal test of prejudicial error—whether it is reasonably probable that a more favorable result would have been reached absent the error (see People v. Watson, supra, 46 Cal.2d 818, 836)—we would conclude that the admission of Siegel’s testimony was reversible error. The judgment is reversed insofar as it relates to penalty. In all other respects the judgment is affirmed. Bird, C.

J., Mosk, J., Newman, J., Reynoso, J., and Grodin, J.,* concurred. Thc present case arises under the 1977 death penalty law (Stats. 1977, eh. 316), repealed by initiative on November 7, 1978. Citations to former Penal Code sections in this opinion refer to the 1977 law. Thc account of the events of April 9 is based upon the testimony at trial, including the recorded statement of defendant to the Kern County Sheriffs Department. We will note those relatively few instances in which the testimony is in conflict.

Defcndant revealed to the psychological experts who examined him that he regularly used PCP and marijuana, and occasionally used other drugs as well. Those statements are admissible, however, only to serve as the basis for expert opinion. (In re Spencer (1965) 63 Cal.2d 400, 412 [46 Cal.Rptr. 753, 406 P.2d 33].) There is no evidence admitted for the truth thereof to show defendant was under the influence of any drug except alcohol. Bufflo’s testimony might imply that Etayo was killed beside the car.

Both defendant and Lufenberger testified, however, that she came out from behind the car and was walking toward defendant when he shot her. The position of the body described by the police witnesses supports the latter version. The trial judge thus anticipated our ruling in Hovey v. Superior Court, supra, 28 Cal.3d 1, 80, that the “portion of the voir dire of each prospective juror which deals with issues which involve death-qualifying the jury should be done individually and in sequestration.” The jury also found that defendant had used a firearm in committing each of the charged offenses. (See Pen.

Code, § 12022.5.) The defense presented three rebuttal witnesses who questioned Rothenberger’s veracity. Defcndant called witnesses at the penally phase who testified to his difficult childhood, his good community reputation for honesty, hard work, and nonviolence, and his relatively good record in county jail awaiting trial. We suggest an additional mitigating factor not mentioned by the trial court: that defendant turned himself in to the police before he was a suspect in the case. Thc public defender stresses that Lufenberger thought defendant was joking and did not take these suggestions seriously.

In light of later events, Lufenberger’s reaction does not prove the suggestion was frivolous, but suggests that Lufenberger did not understand the seriousness of the defendant’s intention. Defendant contends that he received inadequate notice that the prosecution would rely on the felony-murder doctrine. He cites former Penal Code

section 190.4, which requires the People to charge the underlying felorty before asserting the special circumstance of premeditated murder committed in the course of a felony. That provision, however, is inapposite because the People in this case charged only the special circumstance of multiple murder.

With respect to the guilt trial, it has long been settled that under an accusatory pleading charging murder in the short form prescribed by Penal Code sections 951 and 952 the accused may be convicted of first degree murder on the theory that the murder was committed in the perpetration of a felony. (In re Walker (1974) 10 Cal.3d 764, 781 [112 Cal.Rptr. 177, 518 P.2d 1129].) Thus a defendant, charged with murder, is on notice that the prosecution may seek to prove that charge by showing that the homicide occurred during the commission of an enumerated felony.

Defendant points out that an information which complies with sections 951 and 952 may not in some cases give sufficient notice to conform to constitutional requirements. (See People v. Jordan (1971) 19 Cal.App.3d 362, 369 [97 Cal.Rptr. 570; People v. Clenney (1958) 165 Cal.App.2d 241, 253-254 [331 P.2d 696].) He notes also that the California practice of simplified pleading in criminal cases rests largely on the theory that a defendant will learn the details of the charges from the transcript of the testimony before the magistrate or grand jury. (See People v.

Marshall (1957) 48 Cal.2d 394, 399 fn. 5 [309 P.2d 456] and cases there cited.) In the present case the testimony at preliminary hearing, since it did not include testimony by Lufenberger, arguably would not have alerted defendant to the need to defend against a theory of felony murder. Nevertheless defendant, when apprised of the prosecution’s felony-murder theory, did not request a continuance to prepare a defense against that theory; on appeal, defendant does not suggest any way in which the belated injection of the felony-murder theory into this case affected the defense.

We conclude that the prosecution’s failure to give defendant earlier notice of this theory did not impair the presentation of defendant’s defense. The Attorney General also points to Bufflo’s testimony that defendant shouted something to Etayo which included the word “car.” Bufflo could not make out the other words. It is speculative whether defendant said anything which relates to a possible taking of the car. In People v.

Green (1980) 27 Cal.3d 1 [164 Cal.Rptr. 1, 609 P.2d 468], we noted that a murder would not be considered to have been committed in the course of a robbery “when the defendant’s intent is not to steal but to kill and the robbery is merely incidental to the murder.” (P. 61.) In contrast to Green, in which a defendant stole the victim’s clothes and personal effects to conceal her identity, the theft here was not ancillary to the murder, but if Lufenberger’s testimony is believed, a concurrent objective.

As the district attorney argued to the jury in this case, the fact that defendant had an independent intent to kill and in order to kill the victims was willing to risk disabling the car, does not preclude a finding that he also acted with the intent to steal the car. Defendant was questioned earlier by the Norwalk police, but since neither the police nor defendant testified, his statement to the Norwalk police was not offered into evidence. E.g., Greenwald v. Wisconsin (1968) 390 U.S. 519 [20 L.Ed.2d 77, 88 S.Ct. 1152]; Payne v. Arkansas (1958) 356 U.S. 560 [2 L.Ed.2d 975, 78 S.Ct. 844], People v.

Underwood (1964) 61 Cal.2d 113 [37 Cal.Rptr. 313, 389 P.2d 937], also cited by the public defender, involved coercive threats by the interrogating officer. During the Bakersfield interview Deputy Hamilton said to defendant: "... you’ve told us throughout the day that you’re very upset over this thing.... Uh—you and your mind will feel much better if you tell the entire thing and get it off your mind.... [H] Tell us everything you know ... [1]t’ll be a load off your shoulders and I know it’s a load on your shoulders.” Defendant argues that this statement was an improper inducement to confess.

The cases cited by defendant, however, all involve at least an implied promise that defendant will receive better treatment from the police or prosecution if he confesses (see People v. Alfieri, supra, 95 Cal.App.3d 533; People v. Clark (1921) 55 Cal.App. 42 [203 P. 781]); no such implied promise appears in Hamilton’s exhortation. Furthermore, at the time of Hamilton’s statement defendant had already given a statement in Norwalk and repeated the substance of that statement for tape recording in Bakersfield. The officers had no duty to tape record the Norwalk interview. (People v.

Goss (1980) 109 Cal.App.3d 443, 457 [167 Cal.Rptr. 224]; People v. Miller (1975) 52 Cal. App.3d 666, 669-670 [125 Cal.Rptr. 341].) Their failure to do so, however, becomes significant when the officer’s notes, the only other contemporaneous record of the interview, are intentionally destroyed.

Describing defendant's interview with the police, the prosecutor stated that “up to that point the defendant . . . admitted his own conversation about killing people, but never testified, never stated at any time to the officers, at any time earlier, that Lufenberger had suggested that they take their car.” Firsl, in his closing summation, the district attorney said: “Now, if you consider the testimony, and it is uncontradicted concerning the defendant saying that he is going to kill those people, and take the car....” A few minutes later, he told the jury: “You cannot just assume facts that are not in evidence.

You can’t consider things—you have to weigh and consider all this evidence very carefully, and when you weigh that, the testimony is uncontradicted, he went over there with the intention of killing those people and taking their car.” He reiterated “in this case, the testimony is, and it is uncontradicted, he [defendant] said, ‘Let’s kill the people and take their car.’” Finally, the district attorney summed up the prosecution case: “he felt like killing somebody. And he wanted to get a car. Now, he could have gotten a car without shooting them sure.

But because of the attitude which he had, he wanted to kill them. And that, of course, is what makes this case a little unusual. But that doesn’t mean that all the testimony, it did, and there is not testimony that it didn’t happen that way.” The cases cited by the Attorney General are distinguishable. In People v. Jackson (1980) 28 Cal.3d 264 [168 Cal.Rptr. 603, 618 P.2d 149], we characterized the prosecutor’s comment as “indirect, brief, and mild” (p. 305); that comment, in any event, concerned only a minor issue of the case.

Here the prosecutor repeatedly referred to defendant’s failure to contradict the crucial prosecution evidence on the issue of intent and premeditation. In People v. Gaulden (1974) 36 Cal.App.3d 942 [111 Cal.Rptr. 803] and People v. Bethea (1971) 18 Cal.App.3d 930 [96 Cal.Rptr. 229] and People v. Gray (1979) 91 Cal.App.3d 545 [154 Cal.Rptr. 555], the “uncontradicted” testimony could have been contradicted by witnesses other than defendant, so the prosecutor’s comment did not reflect directly on defendant’s failure to testify. Finally, in People v.

Roberts (1975) 51 Cal.App.3d 125 [123 Cal.Rptr. 893], the prosecutorial comment was made to respond to the defense counsel’s reference to “conflicts” in the evidence. The other alleged instances of prosecutorial misconduct are of lesser moment.

(1) The district attorney stated that “1 accept the testimony” of Lufenberger that defendant intended to steal the car. If this rather unclear language was intended to indicate the prosecutor’s personal belief in the truth of that testimony, it was improper. (See People v. Perez (1962) 58 Cal.2d 229, 245 [23 Cal.Rptr. 569, 373 P.2d 617, 3 A.L.R.3d 946] and cases cited.) Taken in context, however, it is unlikely that the statement had any impact on the jury.

(2) The district attorney said that “as a result of [Lufenberger’s] giving of a statement he became near to being prosecuted.” Lufenberger’s statement described him as armed, at the scene of the crime, and doing nothing to prevent it

[…]