OPINION

MOSK, J.

The principal question on this appeal is whether a witness may be allowed to testify after he has undergone hypnosis for the purpose of restoring his memory of the events in issue. The question is new to this court, but has been often litigated in our sister states and extensively studied by medical science. (1a) In accord with recent and persuasive case law and the overwhelming consensus of expert opinion, we conclude that the testimony of such a witness should not be admitted in the courts of California.

I

The record discloses a classic case of conflicting stories. There were only two witnesses to the principal events: the complaining witness, Catherine C., told the jury that defendant compelled her by threat and force to submit to sexual intercourse and to orally copulate him; defendant testified, however, that Catherine willingly participated in the act of intercourse, and there was no oral copulation. The jury believed part of Catherine's story, as it convicted defendant of rape; but it also apparently found that she was lying when she described in detail the alleged act of oral copulation, as it acquitted defendant of that charge. The jury doubtless had a difficult task, since Catherine's performance as a witness was far from exemplary: the record is replete with instances in which her testimony was vague, changeable, self-contradictory, or prone to unexplained lapses of memory. Indeed, on occasion she professed to be unable to remember assertions that she had herself made on the witness stand only the previous day.

In such circumstances it is particularly important that the testimony of the complaining witness be free of taint, lest a mistaken conviction result. Yet as we shall see, in the case at bar the prosecution contaminated Catherine's testimony by subjecting her to a hypnotic experience on the eve of trial for the purpose of "filling the gaps" in her story. To allow her to testify against defendant after that experience was error; and in the light of the entire record, we are of the view that the error caused a miscarriage of justice requiring reversal of the judgment. (Cal. Const., art. VI, § 13.)

A

Catherine was a 32-year-old bartender at a saloon named Bud's Cove, not far from the Camp Pendleton Marine base. The first prosecution witness, Marine Sergeant Charles Lockskin, testified that at 8:50 p.m. on January 25, 1979, he entered Bud's Cove and approached Catherine, whom he had known for several months. She was off duty, and "looked like she was feeling kind of bad." She had a half-consumed martini in front of her, was under the influence of alcohol, and staggered when she walked.

After talking with her for some 15 minutes, Lockskin offered to get her something to eat and take her home. They drove in his car to a take-out restaurant, purchased some food, and arrived at Catherine's apartment house at 9:30 p.m. She vomited when she got out of the car; as this was happening, defendant came up to Lockskin and addressed him by name; Lockskin asked him to leave, and defendant did so. Lockskin then helped Catherine into the apartment and went into the kitchen to prepare some drinks. When he returned to the living room, however, she had passed out on the couch and was fast asleep. After failing to rouse her by shaking her, he covered her with a blanket, turned out the lights, locked the front door, and departed. It was shortly before 10 p.m.

The next witness was Catherine. She testified that on the evening in question she went off duty at Bud's Cove at 6:30 p.m., ordered two martinis, and sat "relaxing" until Lockskin came in. Her testimony as to her activities with Lockskin generally corroborated his, and she admitted she could "feel" the alcohol she had consumed.

Catherine's version of the events occurring after she fell asleep was as follows: she testified that she awoke some time later, still lying on the couch fully clothed, and found defendant standing naked by the coffee table holding a butcher knife. Defendant assertedly took her into the bedroom, ordered her to remove her clothes, and compelled her to orally copulate him for several minutes. The witness admitted that she felt "like I was in a dream" and events were moving in "slow motion."

Catherine then stated that defendant made her get on her knees, tied her hands behind her back and gagged her with nylon stockings, put her head down on the bed, and had intercourse with her in that position for up to half an hour. When she tried to turn her head to see who he was, he struck her with his hand and ordered her not to look at him; later he put a pillow over her head for the same purpose, and struck her on the hip. She claimed the latter blow sobered her so that she no longer felt the effects of her prior drinking.

Until this point the apartment had remained totally dark, and she could see the intruder only as "a shadow." According to Catherine, however, defendant abruptly desisted from further intercourse, removed her bonds and gag, took her back into the living room, and turned on the lights. For the next half hour the two sat naked on the couch, she on his lap, and chatted. Finally he asked her if she liked beer, and she replied that she did; he volunteered to get some from his apartment, and told her where he lived. He dressed and left on this errand; on his return with the beer he took his clothes off again, she got back on his lap, and the conversation resumed.

After another quarter of an hour, defendant suggested they take a shower together, and she agreed. As they entered the bathroom, however, the telephone rang. The caller was assertedly a "girlfriend" of Catherine named Mickie, who announced she was coming over to the apartment. Catherine relayed this fact to defendant, and told him that he could return at another time and she would cook dinner for him. According to Catherine, defendant then got dressed, wrapped the knife and screwdriver in an extra T-shirt he had brought, thrust them down the front of his pants, and left when Mickie arrived. Catherine testified she told Mickie she had been raped by a Marine, and Mickie gave her a strong sedative — a 100-milligram dose of a drug called Mellaril. Mickie stayed for half an hour, and immediately after she left Catherine called the police. According to Catherine, it was 10 minutes before I a.m.

On cross-examination Catherine admitted that during their long conversation in the living room defendant told her numerous personal details about himself, e.g., that he lived in the next apartment building, that his name was Don, that he was 22 years old, that he was married and had a child, that he was a Marine but was not happy in the service, and that the next morning he had to go to Bridgeport, California, for cold-weather training. She claimed that she engaged defendant in the foregoing conversation only because she was afraid he would do her further harm; yet she conceded that when defendant went to get the beer he left the knife and the screwdriver on her living room floor but that she did nothing about them, and that while he was gone she remained sitting naked on the couch. Although she had a telephone she did not call the police or anyone else for help, nor did she dress and go to the nearby apartment of the building manager who was admittedly "a big guy," nor did she even lock the front door. She also acknowledged that she did not know Mickie's last name, address, or telephone number, or where she was at the time of trial, and indeed had never seen her since the night in question.

On redirect examination Catherine testified that until defendant turned on the lights in the apartment, she thought the person having intercourse with her was an older man who resembled defendant and had flirted with her at the bar where she worked.

Police Officer Russell Lane testified that the telephone call reporting the rape came at 1:45 a.m., an hour later than Catherine claimed. He went immediately to her apartment and found her under the influence of alcohol: her breath had the smell of someone who "had been drinking quite heavily," her speech was slow and at times difficult to understand, and her walk was unsteady. She told the officer she had been brought home "very drunk" from Bud's Cove at midnight, that she fell asleep on the couch, and that she awoke in her bed at 12:30 a.m. She gave the officer a physical description of defendant, and repeated the personal information defendant had disclosed to her during their conversation. She then complained that her buttocks hurt, and the officer took her to a local hospital.

At the hospital she was examined by a physician. He testified that he found a bruise on her right hip and "crease marks" on her wrists. But although the latter were consistent with her hands having been tied by a fabric, he could not tell their cause and described them as the kind of marks one receives from sleeping on wrinkles in the bed linen. She reported to the physician that she used "occasional Mellaril and alcohol frequently." He testified that Mellaril is "a major tranquilizer," and that in doses of 100 milligrams or more per day it is prescribed primarily for psychotic states, schizophrenia, and manic-depressive cases.

After the physical examination, Police Officer Leonard Goodwin took a statement of the evening's events from Catherine. The next morning Officer Lane went to defendant's apartment and arrested him as he was leaving to report for duty. When the officer announced the charges were burglary and rape, defendant became angry and said he had "picked up a drunk bitch at Bud's Cove and took her home and fucked her," and "now she wants to report that he raped her" and "that is all a bunch of bullshit."

Defendant took the stand in his own defense. He testified that a few days before these events Catherine had waited on him at Bud's Cove. On the evening in question he entered the bar and saw her sitting with Sergeant Lockskin, whom he recognized. When Lockskin went to the men's room, defendant approached her and asked how she was feeling. They had a brief conversation; according to defendant, she told him her name was Cathy, identified the apartment house in which she lived, and invited him to "grab a six-pack sometime and come over." When Lockskin returned, defendant left the bar and bought some beer at a liquor store. After failing to locate a friend of his, defendant walked to Catherine's apartment house. As he approached, Catherine and Lockskin drove up and defendant spoke briefly with the latter. Defendant then returned to his own apartment for a while, drank some beer, and went back to Catherine's building. When asked why he did so, he explained, "Well, my wife was back home in Indiana. I was by myself. Kind of lonely. And I had an invitation to come to her apartment."

On his arrival, defendant knocked twice on Catherine's door; there was no response, but he thought he heard someone inside who was moaning as if ill. When no one answered further knocking, he called her name through the window and lifted off the screen. He testified that he believed someone inside was sick.

At that point Catherine opened the front door and defendant asked, "Are you okay?" He handed her the screen; she put it next to the front door, went back to the living room, and lay down on the couch. Defendant sat next to her and repeated his question, "Are you okay?" Her reply was to put her arms around his neck and begin kissing him. He responded, and at his suggestion they soon moved to the bedroom. There she cooperated in helping him remove her clothes; defendant returned briefly to the living room for his cigarettes, stripped down, and rejoined her on the bed. They proceeded to have intercourse in the "missionary position," then turned so that he entered her vaginally from behind. She abruptly asked defendant to stop and he did so. He inquired what was wrong, and she replied that she "couldn't be emotionally turned on by men."

Defendant's testimony as to the ensuing events was substantially the same as Catherine's. They sat unclothed on the living room couch talking for half an hour, and he told her all about himself. In turn, she told him that she too was from Indiana, that times were hard for her and she was having problems, and that she had seven children in Knightstown Home for Children. She became upset and began to cry, saying that nobody loved her. As Catherine had testified, defendant went home to get some beer and then suggested they take a shower, but the evening ended when Mickie called on the telephone. He dressed and waited for the latter to arrive, feeling that Catherine "was just in a wrong state of mind to be left alone." After some minutes Mickie entered carrying a six-pack of beer under her arm, and defendant left.

Defendant acknowledged the angry denial he made when Officer Lane accused him of rape the next morning; and he further denied that he broke into Catherine's apartment, or threatened her with a knife or screwdriver, or tied her up or struck her, or had intercourse with her without her consent, or engaged in any act of oral copulation.

Finally, a number of Marine officers, including defendant's platoon commander, his company first sergeant, and his company commander, testified in his behalf. On the basis of their experience they unanimously expressed high personal regard for defendant's truthfulness and honesty, and reported that he had a good reputation for those traits of character. His first sergeant further testified that he was made aware of any altercations occurring in the company, and that defendant had no history of engaging in aggressive or violent behavior.

B

We relate next the evidence bearing on the issue of hypnosis. Prior to trial, counsel for defendant moved to exclude all testimony of the complaining witness that was the result of her having been hypnotized. He offered to prove that the case was originally set for trial on May 1, 1979, but was trailed because of the unavailability of an adequate jury pool; that in the evening of April 30, 1979, i.e., more than three months after the events in question, the deputy district attorney assigned to the case, Richard Fulton, had Catherine hypnotized by another deputy district attorney, Richard Farnell, at the courthouse and in the presence of Mr. Fulton and one Terry Moore; and that Catherine made certain statements under hypnosis which would cause her testimony at trial to be significantly different from her testimony at the preliminary hearing. Counsel then identified one such discrepancy, and argued that "this is an improper use of hypnosis" because "it is not in fact refreshing a witness's recollection" but "it is in fact manufactured evidence." He distinguished those cases in which hypnosis has been used for such purposes as helping an eyewitness to remember a license plate number. He denied that any court in this state had ruled the use of hypnosis permissible in all cases, and charged that here the People were attempting "to expand hypnosis into an area [in] which they cannot lay adequate foundation for its reliability" as a tool for refreshing recollection.

The trial court denied the motion, ruling that prior hypnosis of a witness affects the weight but not the admissibility of the testimony. Accordingly, the court directed that if Catherine gave evidence that she could not remember — or that did not exist — before she was hypnotized, the fact and circumstances of that hypnosis should be put before the jury.

Pursuant to this ruling, Catherine was allowed to testify to a number of matters that she assertedly had been unable to recall on two occasions prior to hypnosis, i.e., when she gave statements to the police on the night of the events in question, and when she testified at the preliminary hearing. For example, on those occasions she stated that after falling asleep in her clothes on the couch in her living room, she awoke in her bedroom and found herself lying naked on the bed, gagged and bound. At trial, as noted above, she testified instead that when she awoke she was still on the couch and fully clothed, and defendant then forced her to go into the bedroom and get undressed. Again, prior to hypnosis she stated that defendant had sexual intercourse with her before as well as after the alleged act of oral copulation, while at trial she testified that the oral copulation preceded any intercourse whatever. Prior to hypnosis she stated that her hands were tied during the oral copulation, while at trial she denied this claim. Finally, prior to hypnosis she stated that the first time she saw the knife in defendant's hand was when they returned to the living room after the sexual intercourse, while at trial she testified she saw it when she awoke on the couch before entering the bedroom.

Both counsel explored the nature and effect of Catherine's hypnotic experience. According to Catherine, before being hypnotized she recalled the events of the evening in question only "vaguely." She discussed the gap in her recollection with Deputy District Attorney Fulton, and consented to be hypnotized "for the purpose of going back over what occurred that night." She verified that she was hypnotized on April 30, 1979, in the courthouse, by Mr. Farnell; although the latter had "some training," he was not a psychiatrist or even a physician. She had not been hypnotized before, but she "just knew" that it enables a person to "remember more than normal."

Apparently she was not disappointed in that expectation. She agreed that the hypnosis at least partly "cured" her recollection as to "this sort of dreamlike period that we're talking about." She credited the hypnosis with causing her to "fill in the gap" in her memory, and also to recall that certain events took place in a different sequence. In particular, she specifically ascribed to the effect of hypnosis each of the above-listed changes between her testimony at trial and her pretrial statements to the police and testimony at the preliminary hearing.

The defense called Dr. Donald W. Schafer as an expert witness to testify on the subject of hypnosis. Dr. Schafer is a board-certified psychiatrist with 16 years of private practice and 10 years on the staff of the University of California at Irvine, where he is a clinical professor of psychiatry. He has had extensive training in hypnosis, and has used it in his practice for two decades. Dr. Schafer acknowledged that hypnosis has certain valid medical uses, such as pain control and relief from various psychosomatic symptoms. In appropriate cases it can also be used for the treatment of neuroses, e.g., by assisting a patient to recover repressed memories of traumatic events, including rape.

Dr. Schafer warned, however, that there are grave risks in relying for other purposes on the accuracy of memories recalled under hypnosis. He explained that while no one knows exactly how the human mind stores information, it does not act like a videotape recorder, i.e., a machine capable of "playing back" the exact images or impressions it has received. Rather, "there are many things that alter the storage of exact memory." There is therefore no assurance, the doctor testified, that a memory recalled in hypnosis is correct. On the contrary, a person under hypnosis can be mistaken in his recollection, or can hallucinate, or can "confabulate," i.e., create a false or pseudomemory, or can even deliberately lie. Indeed, it may be easier to lie under hypnosis, because from the viewpoint of the person in the trance "the hypnosis would put the responsibility on the shoulders of the hypnotist."

Dr. Schafer made four additional important points. First, when a person is put under hypnosis and asked to recount an event, no one is able to determine whether he is telling the truth. Second, when a person has a subconscious motive to distort the truth, e.g., in order to make himself look better in the eyes of others, that motive will usually operate even under hypnosis; indeed, "hypnosis would in a sense give [him] permission" to engage in such distortion. Third, the effect of hypnosis on a preexisting memory is usually additive, i.e., it may permit the recall of additional details; if instead the person remembers the event differently under hypnosis, the discrepancy implies either that his statement describing the preexisting memory was a lie or that the memory under hypnosis was a confabulation. Fourth, when a person has been asked to recall an event while under hypnosis, and after hypnosis is asked to remember the same event, the effect of the prior hypnosis is to remove all doubt he may have had about the event; such persons would be "convinced that what they had said in hypnosis was the truth."

On cross-examination Dr. Schafer testified that although the hypnotic induction in the case at bar was excellent from the viewpoint of technique, the hypnotist did not take into consideration Catherine's possible motivation to distort the truth under hypnosis; one of the factors leading Dr. Schafer to question that motivation was the above-discussed discrepancies in her testimony.

Summing up, Dr. Schafer had no doubt as to the unreliability of hypnosis for discovering the truth of a particular matter. He warned that "hypnosis in no way is a truth serum-like experience," and concluded "there is no way of assessing the reliability of something produced in hypnosis, as such."

The prosecution neither discredited Dr. Schafer's opinion on cross-examination, nor called any expert witness of its own.

II

While passing through periods of vogue and of disrepute, hypnosis has been practiced in one form or another for centuries. Its use in legal proceedings is a relatively recent phenomenon, however, and the rules governing the admissibility of evidence induced by hypnosis are mainly found in the case law of the past two decades. The question of such admissibility has arisen primarily in two contexts: (1) efforts by the defendant to introduce, for the truth of the matter asserted, exculpatory statements made while under hypnosis; and (2) efforts by the prosecution to introduce incriminating testimony of a witness whose memory has assertedly been refreshed by hypnosis. As will appear, the law is well settled as to the former but in a state of flux as to the latter.

A

We unanimously rejected this contention, explaining that "the trial court's ruling did not elevate a fastidious adherence to the technicalities of the law of evidence over the right to a fair trial. For here, unlike Chambers, there was no solid assurance that the hearsay statements were reliable. It appears to be the rule in all jurisdictions in which the matter has been considered that statements made under hypnosis may not be introduced to prove the truth of the matter asserted because the reliability of such statements is questionable. While in California such statements — and those made under the influence of truth serum — may be used to establish a basis for expert opinion, the cases either state specifically or assume that they are not admissible to prove the truth of the matter therein contained. [Citations.]" (Italics added; 25 Cal.3d at p. 665.) We further rejected the defendant's claim that the circumstances of Blair made it likely the witness was telling the truth: "The fact that she was a neutral person and had no reason to falsify her statements under hypnosis and that she intended to tell the truth are obviously insufficient to establish reliability, especially in the light of expert testimony that there is no way to determine if a person under hypnosis is relating actual facts." (Italics added; id. at pp. 665-666.)

As we observed in Blair ( id. at p. 665), "The rule is the same in other jurisdictions." Indeed, no court has held otherwise. Thus in the leading case of Greenfield v. Commonwealth (1974) 214 Va. 710 [ 204 S.E.2d 414, 92 A.L.R.3d 432], a defendant who had no memory of the events surrounding the crime nevertheless made statements relating to those events while hypnotized. In holding the statements inadmissible, the Virginia Supreme Court stressed that "Most experts agree that hypnotic evidence is unreliable because a person under hypnosis can manufacture or invent false statements. [Citations.] A person under a hypnotic trance is also subject to heightened suggestibility. [Citations.]" ( Id. at p. 419.) In subsequently denying habeas corpus relief to the same defendant, the federal district court stated that "the very reason for excluding hypnotic evidence is due to its potential unreliability." ( Greenfield v. Robinson (W.D.Va. 1976) 413 F. Supp. 1113, 1120.) Other courts have rejected hypnotic evidence expressly because of its lack of reliability, while still others have simply declared such evidence inadmissible per se.

Particularly relevant here are the cases that have excluded this evidence on the ground of the well-known Frye rule. ( Frye v. United States (D.C. Cir. 1923) 293 F. 1013.) That rule conditions the admissibility of evidence based on a new scientific method of proof on a showing that the technique has been generally accepted as reliable in the scientific community in which it developed. ( Id. at p. 1014.) Finding that no such showing had been made with regard to hypnosis, the Oklahoma court held in Jones v. State (Crim.App. 1975) 542 P.2d 1316, 1326-1327, that expert testimony as to the truthfulness of statements made by the defendant under hypnosis was inadmissible for the same reason that the results of lie detector and "truth serum" tests are excluded, i.e., because such tests "have not attained sufficient scientific and psychological accuracy nor general recognition as being capable of definite and certain interpretation." ( Id. at p. 1326.)

Again, the Michigan court so held in People v. Hangsleben (1978) 86 Mich. App. 718 [ 273 N.W.2d 539, 543-544], declaring that the defendant's attempt to prove the reliability of statements made in hypnosis by showing the qualifications of the hypnotist "is an inadequate foundation for scientific evidence" under the Michigan version of the Frye rule. ( Id. at p. 544.) And in Rodriguez v. State (Fla.App. 1976) 327 So.2d 903, 904, the Florida court excluded such evidence under its version of Frye, i.e., that the reliability of a new method of proof must be generally accepted by scientists or "'have passed from the stage of experimentation and uncertainty to that of reasonable demonstrability.'" Applying this test, the court held the evidence inadmissible because it was "unconvinced of the reliability of statements procured by way of hypnosis."

B

With this unanimous body of law in mind we turn to the second group of cases mentioned above, i.e., those addressing the admissibility of the testimony of a witness whose memory has assertedly been refreshed by hypnosis. The seminal case was Harding v. State (1968) 5 Md. App. 230 [ 246 A.2d 302]. There the prosecuting witness, Mildred Coley, was found wounded by the roadside, the apparent victim of an aggravated sexual assault; she was in a state of shock and could not remember anything that had happened after being shot by the defendant, who had been riding with her in a car. Several weeks later she was taken to the police barracks to be hypnotized by a psychologist from the state hospital. The police furnished the hypnotist with the details of the case, and he informed Coley he was going to "get her memory back." After he put her under hypnosis he invited two state troopers in, and directed her to tell him "everything that happened" on the day in question. She related certain events incriminating the defendant, with occasional prompting by the hypnotist. He denied suggesting to her any answers to his questions, but did give her the suggestion that after she awoke she would relate the same events. He then brought her out of the trance, and under questioning by one of the state troopers she duly gave the same answers she had given while hypnotized; the hypnotist conceded that his posthypnotic suggestion had made her "want to do so." He testified that in his opinion her story under hypnosis was reliable because certain of her statements were corroborated, because "her recall afterwards was essentially the same," and because she had "no reason" to lie. On the witness stand Coley gave the story a third time, claiming that "When I was asleep it all came back to me."

On appeal from his conviction of assault with intent to commit rape, the defendant urged that the pretrial hypnosis rendered Coley's testimony inadmissible. Affirming the judgment, the Maryland Court of Special Appeals summarily dismissed this contention on the single ground that the witness believed her memory of the events was accurate: "The admissibility of Mildred Coley's testimony concerning the assault with intent to rape case causes no difficulty. On the witness stand she recited the facts and stated that she was doing so from her own recollection. The fact that she had told different stories or had achieved her present knowledge after being hypnotized concerns the question of the weight of the evidence which the trier of facts, in this case the jury, must decide." ( Id. at p. 306.)

In the ensuing 10 years these few sentences spawned a series of similar decisions permitting witnesses to testify to recollections that were assertedly refreshed by pretrial hypnosis. The Attorney General, naturally, relies heavily on those decisions in the case at bar. But an examination of the opinions discloses a significant evolution in the approach of the courts to this issue. In the earlier cases, as in Harding, the courts engaged in little or no analysis of the issue, and merely reiterated the general proposition that the fact of hypnosis "goes to the weight, not the admissibility" of the evidence. If they discussed the point at all, the courts simply noted that the witness believed he was testifying from his own memory and that his credibility could presumably be tested by ordinary cross-examination. (See State v. Jorgensen (1971) 8 Or. App. 1 [ 492 P.2d 312, 315]; Wyller v. Fairchild Hiller Corporation (9th Cir. 1974) 503 F.2d 506, 509-510; Kline v. Ford Motor Co., Inc. (9th Cir. 1975) 523 F.2d 1067, 1069-1070; State v. McQueen (1978) 295 N.C. 96 [ 244 S.E.2d 414, 427]; Clark v. State (Fla.App. 1979) 379 So.2d 372, 375.)

As the decade drew to a close, however, the courts began to take notice of the dangers inherent in using hypnosis for this purpose, and developed increasingly complex procedural "safeguards" in the hope of forestalling those dangers. Thus until 1978 the Ninth Circuit Court of Appeals had applied the Harding rule only in civil cases (see Wyller and Kline, supra); in extending the rule in that year to criminal cases, the court warned: "We are concerned, however, that investigatory use of hypnosis on persons who may later be called upon to testify in court carries a dangerous potential for abuse. Great care must be exercised to insure that statements after hypnosis are the product of the subject's own recollections, rather than of recall tainted by suggestions received while under hypnosis." ( United States v. Adams (9th Cir. 1978) 581 F.2d 193, 198-199.) In a footnote at this point the court proposed several safeguards that it apparently believed would eliminate such "potential for abuse."

In People v. Smrekar (1979) 68 Ill. App.3d 379 [24 Ill.Dec. 707, 385 N.E.2d 848, 853], the Illinois Court of Appeals followed the Harding rule but recognized that "the use of hypnosis is not without problems. 'Asking a patient to recall only real events, or to verify aspects of the material as true or false, reduces but does not remove the element of fantasy' [citation]. The hypnotized subject is also subject to suggestion by the hypnotist." Accordingly, the court held admissible the identification testimony of a previously hypnotized eyewitness only because of a number of factors in the record which the court impliedly viewed as guaranteeing reliability.

In subsequent cases the required safeguards became very elaborate indeed. Thus in State v. Hurd (1981) 86 N.J. 525 [ 432 A.2d 86], the eyewitness-victim was unable or unwilling to identify the defendant as her assailant, and did so only after being hypnotized three weeks later. The defendant moved before trial to suppress her proposed in-court identification, and extensive expert testimony was taken on the subject of the reliability of hypnotically induced recollection. The trial court ordered the testimony suppressed.

On appeal, the New Jersey Supreme Court held that the admissibility of hypnotically induced testimony must be judged by the Frye standard, but immediately qualified the rule to require only that in any case the hypnosis produce a recall that is, in effect, no more inaccurate than that of the average witness who has not been hypnotized. ( Id. at pp. 91-92.) The court recognized a number of dangers inherent in the hypnotic process which "explain why hypnosis, unless carefully controlled, is not generally accepted as a reliable means of obtaining accurate recall" ( id. at p. 93); the dangers included the subject's extreme suggestibility, loss of critical judgment, tendency to confabulate, and excessive confidence in his new "memories." The court nevertheless declined to hold such testimony inadmissible per se, asserting that "the reliability of ordinary eyewitnesses reveals similar shortcomings." ( Id. at p. 94.)

Yet to minimize if possible the admitted risks of hypnosis, the court went on to adopt an intricate set of procedural prerequisites to its use. First, the trial court should "evaluate both the kind of memory loss that hypnosis was used to restore and the specific technique employed, based on expert testimony presented by the parties." ( Id. at p. 95.) The court should then inquire into "the amenability of the subject to hypnosis," because persons capable of entering deeper trances may be more suggestible. ( Id. at p. 96.) In turn, the party offering the testimony must prove he has complied with no less than six additional procedural requirements, intended to furnish an adequate record and insure "a minimal level of reliability." ( Id. at pp. 96-97.)

Finally, in order to guarantee "strict compliance" with these prerequisites, the proponent of the testimony must establish its admissibility by "clear and convincing" proof. ( Id. at p. 97.) "This burden," said the New Jersey court, "is justified by the potential for abuse of hypnosis, the genuine likelihood of suggestiveness and error, and the consequent risk of injustice." ( Ibid.) Because the court found that several of the listed procedural requirements had not been met on the record in Hurd, it affirmed the order suppressing the proposed testimony.

C

Lastly, even if requirements could be devised that were adequate in theory, we have grave doubts that they could be administered in practice without injecting undue delay and confusion into the judicial process. To be sure, it would usually be easy to determine if the hypnotist was an appropriately trained psychiatrist or psychologist. It might be harder to establish that he was sufficiently independent of the prosecution or defense to avoid subconscious bias. And it would certainly be far more difficult to prove strict compliance — which Hurd demands — with each of the remaining "safeguards." It strains credulity, for example, to believe that a conscientious defense counsel would meekly agree that the prosecution had recorded every bit of relevant information conveyed to the hypnotist prior to the session, or that the hypnotist had conveyed absolutely none of that information to the subject either while extracting the latter's prehypnotic version of the facts or while questioning him both during and after hypnosis, or that every single contact between the hypnotist and the subject, no matter how innocuous, had been preserved on videotape.

On the other hand, it takes little prescience to foresee that these and related issues would provide a fertile new field for litigation. There would first be elaborate demands for discovery, parades of expert witnesses, and special pretrial hearings, all with concomitant delays and expense. Among the questions our trial courts would then be expected to answer are scientific issues so subtle as to confound the experts. (See, e.g., fn. 24, ante.) Their resolution would in turn generate a panoply of new claims that could be raised on appeal, including difficult questions of compliance with the "clear and convincing" standard of proof. And because the hypnotized subject would frequently be the victim, the eyewitness, or a similar source of crucial testimony against the defendant, any errors in ruling on the admissibility of such testimony could easily jeopardize otherwise unimpeachable judgments of conviction. In our opinion, the game is not worth the candle.

For all these reasons, we join instead a growing number of courts that have abandoned any pretense of devising workable "safeguards" and have simply held that hypnotically induced testimony is so widely viewed as unreliable that it is inadmissible under the Frye test. This disposition, of course, is consistent with the above-discussed case law uniformly excluding evidence of the truth of statements made under hypnosis. (See pt. II A, ante.) And both rules, as we shall see, are supported by the overwhelming consensus of contemporary scientific opinion on hypnosis.

The first case to depart from the Harding approach was People v. Hangsleben (1978) supra, 273 N.W.2d 539. There, after he had confessed, the defendant was hypnotized and assertedly recalled the events of the crime for the first time, stating that a third party was the true culprit. He sought to prove at trial that the hypnosis had refreshed his recollection, in order to bolster his story on the witness stand and to explain his prior inconsistent admission to the police. The Michigan court distinguished Harding, and held that the evidence was properly excluded because the defendant "failed to establish the reliability of hypnosis as a memory-jogging device." ( Id. at p. 544.) His sole showing was to assert the qualifications of the hypnotist and to refer to the theory of memory restoration by hypnosis. Ruling "That does not demonstrate the general scientific acceptance" required by the Michigan version of the Frye test, the court rejected the evidence for lack of proof that hypnosis has been successful in restoring the memory of others, either by their testimony or that of experts. ( Id. at p. 545.)

In State v. La Mountain (1980) 125 Ariz. 547 [ 611 P.2d 551], it was the prosecution that failed to prove the reliability of hypnosis used to restore a witness' recollection. The defendant was convicted of sexually assaulting a customer in a laundromat. At trial, two prosecution witnesses identified the defendant as the person who committed a similar assault in the same laundromat fifteen months earlier; one was the victim of that assault, and the other was a bystander who seized the assailant. Both witnesses, however, had been unable to identify the defendant from a photographic lineup until their memories were "refreshed" by hypnosis. The hypnotist was a deputy sheriff who had attended various law enforcement institutions giving instruction in hypnotism. He used a so-called "TV technique," asking the subject to visualize the events of the crime as if they were being played back on a videotape machine.

On appeal, the Arizona Supreme Court held it was error to allow the two witnesses to the prior assault to testify after they had been hypnotized. The court reasoned (at p. 555), "There was no expert testimony regarding the effect of hypnosis upon a person's memory, and we do not know from the record what effect the previous hypnotic identification had on the witness's later in-court testimony and identification. Although we perceive that hypnosis is a useful tool in the investigative stage, we do not feel the state of the science (or art) has been shown to be such as to admit testimony which may have been developed as a result of hypnosis. A witness who has been under hypnosis, as in the case here, should not be allowed to testify when there is a question that the testimony may have been produced by that hypnosis." The court nevertheless affirmed the judgment, finding from the evidence that the result would have been the same if these witnesses had not testified.

The gap in proof identified in Hangsleben and La Mountain was quickly and thoroughly filled in the leading case of State v. Mack (Minn. 1980) 292 N.W.2d 764. In that case the defendant met Marion Erickson in a bar, and eventually took her to a motel on his motorcycle. Thereafter he telephoned for an ambulance and told the drivers that he and Erickson had been engaged in intercourse when she started bleeding from the vagina. Erickson was drunk, her speech was unclear, and she had difficulty walking. At the hospital a single deep cut was found inside her vagina; she told one intern that fingers had been inserted in her vagina during sexual activity, and another that she had been in a motorcycle accident. After the doctors advised her they did not believe her explanations, she reported to the police that she had been assaulted. She could not, however, remember much of the events of the night in question.

Six weeks later the police caused Erickson to be hypnotized by a lay hypnotist without professional training. After placing her in a deep trance, the hypnotist invited the police investigator and another officer to join them. He then told Erickson that she would recall the events as though on a television screen. In the course of the session Erickson accused the defendant of stabbing her repeatedly in the vagina. At the end of the session the hypnotist gave her a posthypnotic suggestion to the effect that she would be able to remember very clearly everything that happened on the night in question. The next day she gave the same police investigator a written statement recounting as her present memory the events she had related under hypnosis.

The defendant was arrested and charged with aggravated sexual assault. The question of the admissibility of Erickson's proposed testimony was litigated at an extensive pretrial hearing. Following that hearing, and pursuant to Minnesota procedure, the trial court stayed the prosecution and certified to the state supreme court the question "whether a previously hypnotized witness may testify in a criminal proceeding concerning the subject matter adduced at the pretrial hypnotic interview." (292 N.W.2d at p. 765.) In a well-reasoned opinion the Minnesota high court unanimously answered in the negative, holding such testimony inadmissible as a matter of law.

The court began by emphasizing that no less than five experts on hypnosis and memory retrieval had testified at the hearing, making an extensive record on which to decide the legal issue. The court also observed that the record "demonstrates the truth of Dr. Orne's observation that a case-by-case decision on the admissibility question would be prohibitively expensive, and reveals the difficulty of getting experts qualified to testify about hypnosis as an investigative rather than a therapeutic tool." ( Id. at p. 766.)

The defendant in Mack contended that Erickson's hypnotically refreshed recollection was too unreliable to merit admission, and that to allow such testimony would deny him the right to effective cross-examination. The state contended that the testimony should be admitted as long as certain "safeguards" can be established, relying on Harding and its progeny; the defendant, in turn, invoked Frye. The state argued that Frye is inapplicable to evidence that is not the direct product of a mechanical device such as a lie detector; and to be admissible, the testimony of a previously hypnotized witness need not be true provided it is based on what the witness actually perceived. Stressing the potentially drastic effect of hypnosis on a witness' testimony, however, the court ruled that "Although hypnotically-adduced 'memory' is not strictly analogous to the results of mechanical testing, we are persuaded that the Frye rule is equally applicable in this context" ( id. at p. 768).

The court turned to the record to determine whether, under Frye, the use of hypnosis to refresh a witness' memory has been generally accepted as reliable by the scientists working in the field. The court found that the exact opposite is true, i.e., that the consensus of informed expert opinion rejects the use of hypnosis for this purpose because it is "not scientifically reliable as accurate." ( Id. at p. 768.)

The court gave a number of reasons for this conclusion, each drawn from the expert testimony before it. ( Id. at p. 768.) First, "a hypnotized subject is highly susceptible to suggestion, even that which is subtle and unintended. Such suggestion may be transmitted either during the hypnotic session or before it," by such persons as police officers or doctors. This suggestibility is enhanced by the subject's natural "desire to please either the hypnotist or others who have asked the person hypnotized to remember and who have urged that it is important that he or she remember certain events." The result is that "hypnosis can create a memory of perceptions which neither were nor could have been made" ( id. at p. 769). And "Most significantly, there is no way to determine from the content of the 'memory' itself which parts of it are historically accurate, which are entirely fanciful, and which are lies." ( Id. at pp. 768-769.)

The expert testimony also supported the defendant's claim of denial of effective cross-examination: "In addition to its historical unreliability, a 'memory' produced under hypnosis becomes hardened in the subject's mind. A witness who was unclear about his 'story' before the hypnotic session becomes convinced of the absolute truth of the account he made while under hypnosis. This conviction is so firm that the ordinary 'indicia of reliability' are completely erased, . . . It would be impossible to cross-examine such a witness in any meaningful way." (Fn. omitted; id. at p. 769.)

Summing up, the court recognized but declined to perpetuate the two inconsistent lines of cases discussed hereinabove (pts. II A and II B, ante): "We follow the best scientific authority, however, in rejecting as artificial and unprincipled any distinction between hypnotically-induced testimony offered by the defense to exculpate and that offered by the prosecution to make its case. Regardless of whether such evidence is offered by the defense or by the prosecution, a witness whose memory has been 'revived' under hypnosis ordinarily must not be permitted to testify in a criminal proceeding to matters which he or she 'remembered' under hypnosis." ( Id. at p. 771.)

Because the precise question certified to the Minnesota Supreme Court in Mack was limited to the admissibility of a previously hypnotized witness' testimony on the subject matter of the hypnotic session, the court did not resolve the larger question whether such a witness should be allowed to testify on other matters relating to the crime that were not expressly covered in the hypnotic session and were allegedly recalled without the aid of hypnosis. The latter question was soon answered, again in the negative.

In People v. Tait (1980) 99 Mich. App. 19 [ 297 N.W.2d 853], the defendant was charged with assaulting Deputy Sheriff Myers with intent to commit murder. At the preliminary hearing Myers testified that the defendant approached him, raised a pistol, and twice threatened to blow his head off; when the defendant ignored his orders to stop and came nearer, Myers shot him. At trial Myers told the same story, with one difference: whereas at the preliminary hearing he had testified that he did not see the defendant attempt to fire the pistol, at trial he testified that just before he shot the defendant he saw the latter move his hand to the top of the weapon. Defense counsel moved for a mistrial, stating he had not learned that Myers' memory had been refreshed by pretrial hypnosis until he so testified. The hypnotist had been the prosecuting attorney. Myers claimed that at the hypnotic session no one told him what to say, and that his trial testimony was his own recollection of the incident. The trial court denied the motion, and the defendant was convicted.

Reversing the judgment, the Michigan court began by recalling its 1978 decision in Hangsleben, supra, which held similar testimony inadmissible. The People argued that the Frye rule was inapplicable in this context because the witness' recollection was merely "refreshed" as permitted by law. The court rejected this contention, warning that "Investigatory use of hypnosis on persons who are later called on to testify in court carries a dangerous potential for abuse." ( Id. at p. 856.) The court then held the case governed by the Michigan version of the Frye rule, to wit, that "general scientific recognition [must] be established by testimony of disinterested and impartial experts or disinterested scientists whose livelihood was not intimately connected with the technique. [¶] In the instant case the technique is not new, but we believe the same requirements must be met as are required for the introduction of lie detector or voicewriter evidence or evidence influenced by them." ( Id. at p. 857.) Applying that test, the court held ( ibid.) that "Hypnosis has not 'achieved that degree of general scientific acceptance' which will permit its introduction," citing inter alia our decision in People v. Kelly (1976) 17 Cal.3d 24 [ 130 Cal.Rptr. 144, 549 P.2d 1240].

The question of disposition remained. The Frye error, noted the court, was compounded by the prosecution's failure to disclose before trial the fact that the witness had been hypnotized to restore his memory. The court then concluded, "By virtue of the prosecutor's improper actions in this case, deputy sheriff Kirk Myers' testimony has been damaged to the extent that it cannot be used on retrial of the case. In lieu of discharge, the case is remanded for retrial, but the prosecution shall be absolutely prohibited from in any way using any testimony of deputy sheriff Myers. The trial court is adjured to permit no testimony of any kind as to what Myers may have seen or heard." (Italics added; 297 N.W.2d at p. 857.)

Any inference, however, that prosecutorial impropriety is a precondition to exclusion of the testimony of a previously hypnotized witness was firmly dispelled in the recent Arizona case of State v. Mena. There Stephen Koors was stabbed outside a bar by three men. Sometime later he went with a police officer to see two doctors who hypnotized him for the purpose of improving his memory of the assault. The doctors questioned him about the incident, and his answers contained more details than the statements he had previously given to the police. Before terminating the session the doctors told him that after he came out of hypnosis he would remember what he had related to them in the trance. At trial, the defendant moved to exclude Koors' testimony unless it could be shown that his memories of the event were his own recollection and not implanted by suggestions of the hypnotists. The motion was denied and the defendant was convicted.

At the first level of review the Arizona Court of Appeal affirmed the judgment, relying on the "respectable authority" of Harding and its progeny to find no fundamental error. ( State v. Mena (App. 1980) 128 Ariz. 244 [ 624 P.2d 1292, 1294].) Declining in effect to apply the Frye rule, the court reasoned that "hypnotically adduced evidence cannot be equated with, for example, the results of a lie detector examination since one can cross-examine the witness but cannot cross-examine the lie detector." ( Ibid.)

On further review, however, the Arizona Supreme Court rejected both the authority and the reasoning of the court of appeal, vacated the portion of its opinion dealing with hypnosis, and reversed the assault conviction. ( State v. Mena (1981) 128 Ariz. 226 [ 624 P.2d 1274].) The heart of its decision is a careful inquiry into the scientific realities of hypnosis and its effect on potential witnesses. In that inquiry the court relied on a number of scholarly articles as support for conclusions identical to those drawn in Mack from expert testimony: i.e., persons under hypnosis are prone to experience false memories, fantasies, and confabulations; these distortions are aggravated by the subject's tendency to respond in the way he believes the hypnotist desires, even without the awareness of either; the subject is unable to distinguish his true memories from pseudomemories implanted during hypnosis; and after hypnosis he will often be more convinced of the accuracy of the latter than the former, making cross-examination ineffective. ( Id. at pp. 1276-1278.) On these grounds the court discredited Harding and its progeny, and instead quoted with approval from the above-discussed opinion in Tait.

The court then reiterated the Arizona version of the Frye rule (citing inter alia People v. Kelly (1976) supra, 17 Cal.3d 24), and stated, "We believe that the same standard should apply to the use of hypnosis to produce testimony by purportedly improving memory." (624 P.2d at p. 1279.) Applying the rule in this context, the court found that although hypnosis has certain approved therapeutic uses, its use "to aid in accurate memory recall is not yet generally accepted." ( Ibid.) Reaffirming the holding of its 1980 decision in La Mountain, supra, the court therefore concluded ( ibid.): "The determination of the guilt or innocence of an accused should not depend on the unknown consequences of a procedure concededly used for the purpose of changing in some way a witness' memory. Therefore, until hypnosis gains general acceptance in the fields of medicine and psychiatry as a method by which memories are accurately improved without undue danger of distortion, delusion or fantasy, we feel that testimony of witnesses which has been tainted by hypnosis should be excluded in criminal cases."

In the course of its opinion the court quoted the published view of Dr. Bernard L. Diamond that a witness who has been hypnotized for the purpose of improving his memory is so contaminated that he is thereafter incompetent to testify. ( Id. at p. 1277.) In its disposition the court adopted that view and the corresponding holding of Tait, ruling that after Koors had been hypnotized it was prejudicial error to allow him to testify at all. The court explained (at p. 1280) that "it will often be difficult to determine whether proferred testimony has been produced by hypnosis or has come from the witness' own memory, unaffected by hypnotic suggestion. In order to ensure against the dangers of hypnosis, therefore, this Court will consider testimony from witnesses who have been questioned under hypnosis regarding the subject of their offered testimony to be inadmissible in criminal trials from the time of the hypnotic session forward." (Italics added.) In a footnote at this point the court recognized that "our decision today may place the state in the difficult position of choosing whether to use a particular witness' testimony at a criminal trial or to subject that witness to hypnotism as an investigatory tool. We do not pass at this time on the state's ability to preserve a witness' prehypnotic testimony" by means of deposition.

With the next case in this series, we come full circle. As we have seen, the sequence began in 1968 with the decision of the Maryland Court of Special Appeals in Harding; but in Polk v. State (1981) 48 Md. App. 382, [ 427 A.2d 1041], the same court has now repudiated Harding and held instead that Frye governs the question before us. This dramatic turn of events would appear to give the coup de grace to the moribund precedent relied on here by the Attorney General.

In Polk the defendant was charged with orally copulating an eight-year-old boy named Bobby, his next-door neighbor. Some five months after the incident Bobby was taken to the state police barracks to be hypnotized by Sergeant White, a police investigator with minimal experience in hypnosis. Before the session began White was advised that the goal was to determine whether Bobby could remember any sexual contact with this defendant. Bobby's mother and the prosecuting attorney were present during the session. Under White's questioning, Bobby produced details of the alleged misconduct. The defendant moved before trial to suppress the testimony of Bobby and White on the ground that it would be the product of "an inexact and unproven science" and hence was inadmissible as a matter of law. Counsel also contended that White was not qualified as a hypnosis expert, his questions to Bobby were improperly suggestive, and there was an impermissible delay between the incident and the hypnotic session. The trial court denied the motion, asserting that the fact of hypnosis goes to the weight rather than the admissibility of the testimony. This, of course, was the rule in Maryland since Harding.

Bobby's testimony at trial was substantially the same as the answers he gave under hypnosis. On cross-examination he stated he had no recollection of the incident and "forgot the nasty part" until he spoke with White. The prosecution did not call White, however, and did not introduce evidence of the hypnosis. For that reason the court refused to allow a defense expert to testify that hypnotically refreshed recollection is unreliable and White's hypnotic procedure was improper. The defendant's motions to strike Bobby's testimony and for acquittal or mistrial were denied, and he was convicted.

The appellate court acknowledged that Bobby's testimony would have been admissible under its 1968 decision in Harding, but held that decision had been undermined 10 years later when Maryland first adopted the Frye rule in the case of Reed v. State (1978) 283 Md. 374 [ 391 A.2d 364, 97 A.L.R.3d 201]. The court in effect distinguished away Harding on the following rationale: "In Harding, we did not assess the Frye principle, the rule there enunciated not having been applied in this State until Reed; nor did we have occasion to probe the question — here directly raised on the authority of Reed — of the general acceptability of hypnotism as a reliable technique for memory retrieval within the relevant scientific community." (Fns. omitted; 427 A.2d at p. 1047.)

To answer that question, the court quoted at length from both Mack and Mena (and cited Tait as being in accord), stressing their reliance on the Frye rule in judging the admissibility of hypnotically refreshed recollection. The court then reasoned ( id. at p. 1048): " Reed speaks, of course, of expert testimony based upon the use of a scientific technique. The technique of hypnosis is scientific, but the testimony itself of the witness is the end product of the administration of the technique. The induced recall of the witness is dependent upon, and cannot be disassociated from, the underlying scientific method. Accordingly, we conclude, as did the Minnesota and Arizona Courts, that the Frye test must be applied in the instant case, i.e., before Bobby's testimony can be admitted, there must be a determination of whether hypnosis is generally acceptable in the relevant scientific community for the purpose of memory retrieval."

Because that determination had been precluded below by the refusal to allow the defendant to prove the unreliability of hypnosis, the court reversed and remanded with directions to the trial judge to rule in the first instance on the general acceptability of this technique as shown by expert testimony and scholarly publications. Finally, adopting the Mena rule of total disqualification of any witness thus contaminated, the court also directed that if the technique is found inadmissible under Frye, Bobby must not be allowed to testify at all on the retrial "since the boy had no recollection of the alleged incident giving rise to the charges against the appellant prior to the hypnosis. State v. Mena, supra." ( Id., 427 A.2d at p. 1049.)

In Com. v. Nazarovitch (1981) 496 Pa. 97 [ 436 A.2d 170], the Supreme Court of Pennsylvania not only joined the foregoing line of decisions barring hypnotically induced testimony, but expressly refused to follow Hurd in its attempt to sanitize such evidence by procedural "safeguards." The defendants in Nazarovitch were charged with murder on the basis of statements by a witness whom the police had caused to be hypnotized on three occasions prior to trial in an effort to refresh her memory of the events. Law enforcement officials were present during each hypnotic session, and furnished some of the questions asked of the witness. The defendants' pretrial motion to suppress the witness' testimony was granted, and the prosecution appealed.

In a unanimous opinion, the Pennsylvania high court held such testimony inadmissible and affirmed the suppression order. The court began by ruling that the admissibility of the challenged testimony must be judged by the Frye test, i.e., "whether hypnotically-refreshed testimony is generally accepted in the scientific community as yielding reasonably reliable results." ( Id. at p. 173.) As in Mena, the court answered the question by examining the published views of leading representatives of that community. From such studies the court concluded, as in both Mena and Mack, that the scientific community has "grave misgivings" about the reliability of hypnosis in forensic use, for a number of reasons inherent in the phenomenon itself: i.e., "the heightened suggestivity, the increased desire to satisfy the hypnotist, the tendency to confabulate, and the inability to distinguish in one's waking state the fact from the fantasy" ( id. at p. 174). "Furthermore," the court observed, "the hypnotic subject, upon awakening, is often imbued with a confidence and conviction as to his memory which was not present before. Prehypnosis uncertainty becomes molded, in light of additional recall experienced under hypnosis, into certitude, with the subject unaware of any suggestions that he acted upon or any confabulation in which he engaged. The subject's firm belief in the veracity of his enhanced recollection is honestly held, and cannot be undermined through cross-examination." ( Ibid.)

The court then carefully reviewed the reasoning and holding of its sister state in Hurd, but refused to accede to the urging of the prosecution that it adopt such an approach in Pennsylvania. Rather, the court explained, "we remain unconvinced that the trier of fact could do anything more than speculate as to the accuracy and reliability of hypnotically-refreshed memory. The Hurd court's rationale that hypnotically-refreshed recollection might as well be admissible since ordinary eyewitness accounts are also vulnerable to error and inaccuracies does not do full justice to the fact that 'the traditional guaranties of trustworthiness as well as the jury's ability to view the demeanor of the witness are wholly ineffective to reveal distortions of memory induced by the hypnotic process.' [Citation.] It is unchallenged that a jury can more critically analyze a witness' ability to perceive, remember, and articulate his recollections when such testimony has not been hypnotically-refreshed. The probative worth of the hypnotically-adduced evidence cannot overcome the serious and fundamental handicaps inherent therein." ( Id. at pp. 176-177; accord, State v. Palmer (1981) 210 Neb. 206 [ 313 N.W.2d 648, 653-655] [rejecting Hurd and holding hypnotically induced testimony inadmissible for lack of general scientific acceptance]; People v. Gonzales (1981) 108 Mich. App. 145 [ 310 N.W.2d 306, 308-314] [same].)

III

From the foregoing cases it appears that the correct analysis of the problem before us is to determine whether hypnotically recalled testimony is subject to the California version of the Frye rule, and if so, whether it meets the test of that rule. We proceed to such an analysis.

A

The Frye rule is deeply ingrained in the law of this state. It has repeatedly been invoked by our courts to determine the admissibility of evidence based, for example, on polygraph examinations ( People v. Wochnick (1950) 98 Cal.App.2d 124, 127-128 [ 219 P.2d 70]), "truth serum" ( People v. Jones (1959) 52 Cal.2d 636, 653 [ 343 P.2d 577]), Nalline testing ( People v. Williams (1958) 164 Cal.App.2d Supp. 858, 860-862 [331 P.2d 251]), experimental systems of blood typing ( Huntingdon v. Crowley (1966) 64 Cal.2d 647, 653-656 [ 51 Cal.Rptr. 254, 414 P.2d 382]), voiceprint analysis ( People v. Kelly (1976) supra, 17 Cal.3d 24), human bite marks ( People v. Slone (1978) 76 Cal.App.3d 611, 623-625 [ 143 Cal.Rptr. 61]), and microscopic identification of gunshot residue particles ( People v. Palmer (1978) 80 Cal.App.3d 239, 250-255 [ 145 Cal.Rptr. 466, 1 A.L.R.4th 1056]). We recently reviewed the several reasons for this rule in Kelly (17 Cal.3d at pp. 31-32), and need not repeat them here; it is enough to note our conclusion ( id. at p. 32) that "we are persuaded by the wisdom of, and reaffirm our allegiance to, the Frye decision and the 'general acceptance' rule which that case mandates."

The Attorney General contends the Frye rule is inapplicable in the present context, making in essence the following argument: The rule is assertedly limited to cases in which (1) an expert witness gives his opinion (2) interpeting the results of a new technique for scientifically testing or analyzing physical evidence, and (3) that opinion goes directly to the existence or nonexistence of a disputed fact, which is often the ultimate issue in the litigation. By contrast, in cases such as the present it is not the expert (i.e., the hypnotist) who ordinarily testifies; the process involved (i.e., the hypnotizing of a potential witness to improve his recall) has nothing to do with testing physical evidence; if the expert does testify, he should not be asked to interpret the results of the technique (i.e., to give his opinion on whether the revived memories of the hypnotized subject are true) but simply to discuss its methodology (i.e., to explain how the hypnotic session was conducted); and the latter testimony evidently does not go to the disputed fact or ultimate issue (e.g., the identity of the culprit). Rather, in the typical case the witness is the person who actually perceived the event that is the subject of the litigation, and his testimony is the same as that of any other lay witness, i.e., he states his present recollection of that event to the best of his ability. It is true that his recollection has been refreshed by hypnosis, and that hypnosis does not guarantee truthful or accurate recall. But neither does any other method of reviving memory. That guarantee, as with all witnesses, comes from cross-examination, which permits the trier of fact to determine the truth and accuracy of the hypnotically refreshed testimony.

The argument is unpersuasive for a number of reasons. First, it proceeds from an unduly narrow reading of the opinions invoking the Frye rule: as we said in Kelly, for example, the rule applies to evidence "developed by" or "based upon" new scientific techniques. (17 Cal.3d at p. 31.) Nor are those techniques necessarily limited to manipulation of physical evidence: we do not doubt that if testimony based on a new scientific process operating on purely psychological evidence were to be offered in our courts, it would likewise be subjected to the Frye standard of admissibility. In either case, the rule serves its salutary purpose of preventing the jury from being misled by unproven and ultimately unsound scientific methods. ( Kelly, at pp. 31-32.)

Moreover, from the unchallenged expert testimony in the case at bar and the uniform findings of the jurisdictions that have inquired into the matter, it appears that hypnotizing a witness to improve his memory is not in fact like "any other method" of refreshing a witness' recollection. These sources reveal that the hypnotic process does more than permit the witness to retrieve real but repressed memories; it actively contributes to the formation of pseudomemories, to the witness' abiding belief in their veracity, and to the inability of the witness (or anyone else) to distinguish between the two. In these circumstances, as noted above, the resulting recall of the witness "is dependent upon, and cannot be disassociated from" the underlying hypnosis. ( Polk v. State (1981) supra, 427 A.2d at p. 1048.) And if the testimony is thus only as reliable as the hypnotic process itself, it must be judged by the same standards of admissibility.

The question has in any event been decided in California. In People v. Diggs (1980) 112 Cal.App.3d 522 [ 169 Cal.Rptr. 386], the court held Kelly applicable in this context on essentially the same reasoning: "In view of the modification of memory and demeanor which generally follow from treatment by hypnosis, we are persuaded that posthypnotic testimony may in many instances properly be termed a product of the technique. The admissibility of evidence based upon a new scientific technique is governed by People v. Kelly. . . . [¶] The Kelly court was concerned with and sought to mitigate the dangerous tendency of lay jurors to give considerable and often undue weight to scientific evidence presented by experts with impressive credentials. (17 Cal.3d, at p. 31.) Such procedures are invested with a '"misleading aura of certainty which often envelops a new scientific process, obscuring its currently experimental nature."' (17 Cal.3d at p. 32.) This is certainly true of hypnosis." (112 Cal.App.3d at pp. 530-531.) (2) (See fn. 32.) We approve this portion of Diggs. (1b) In accord, therefore, with the courts of Michigan, Minnesota, Arizona, Maryland, and Pennsylvania, we hold that in this state the testimony of witnesses who have undergone hypnosis for the purpose of restoring their memory of the events in issue cannot be received in evidence unless it satisfies the Frye standard of admissibility.

B

It is the proponent of such testimony, of course, who has the burden of making the necessary showing of compliance with Frye, i.e., of demonstrating by means of qualified and disinterested experts that the new technique is generally accepted as reliable in the relevant scientific community. ( Kelly, at pp. 36-40.) In the case at bar the prosecution did not take up that burden, and made no such showing. On this ground alone we would ordinarily be justified in holding the challenged testimony inadmissible. But while such a ruling would dispose of the appeal before us, it would provide the bench and bar with little guidance in other litigation presenting the same questions. Moreover, in the particular circumstances of this case the prosecution had little if any opportunity to make the required showing. We therefore reach the Frye issue on its merits, both for reasons of precedent and considerations of fairness.

(3) Yet it does not follow, as the Attorney General claims, that the record is inadequate to support a decision on the general admissibility of hypnotically aided testimony. Here the issue was fully raised by defendant. Not only did he make a timely pretrial motion on this ground, but on cross-examination of Catherine he probed critically into the purpose, method, and results of her hypnotic experience. By the testimony of Dr. Schafer, defendant then exposed the multiple risks in using hypnosis to restore a witness' memory; and he elicited from Dr. Schafer an unequivocal expert opinion that hypnosis is not reliable as a truth-seeking technique.

To be sure, in Kelly we doubted whether the testimony of a single witness, even if qualified, is sufficient to establish the views of an entire scientific community as to the reliability of a new procedure. (17 Cal.3d at p. 37.) As will appear, however, Dr. Schafer's testimony is supported by a substantial body of scholarly treatises and articles on the subject. The Attorney General complains that "literature is not evidence," and that it would be improper for this court to "pick and choose among that literature to decide issues of scientific fact." The remark betrays a fundamental misunderstanding of the task before us: our duty is not to decide whether hypnotically induced recall of witnesses is reliable as a matter of "scientific fact," but simply whether it is generally accepted as reliable by the relevant scientific community. We recognized in Kelly ( ibid.) that "Ideally, resolution of the general acceptance issue would require consideration of the views of a typical cross-section of the scientific community, including representatives, if there are such, of those who oppose or question the new technique." But considerations of judicial economy make it impractical to require those views to be presented personally by each scientist testifying in open court: as pointed out in Mack (292 N.W.2d at p. 766), such a procedure would be prohibitively expensive, and would be frustrated in any event by the difficulty of finding local experts qualified to testify on hypnosis as an investigative rather than a therapeutic tool.

Accordingly, for this limited purpose scientists have long been permitted to speak to the courts through their published writings in scholarly treatises and journals. ( Kelly, at p. 35; Huntingdon v. Crowley (1966) supra, 64 Cal.2d 647, 656; People v. Palmer (1978) supra, 80 Cal.App.3d 239, 252-254; People v. Law (1974) 40 Cal.App.3d 69, 75 [ 114 Cal.Rptr. 708]; United States v. Addison (D.C. Cir. 1974) 498 F.2d 741, 744-745.) The courts view such writings as "evidence," not of the actual reliability of the new scientific technique, but of its acceptance vel non in the scientific community. Nor do the courts "pick and choose" among the writings for this purpose. On many topics — including hypnosis — the scientific literature is so vast that no court could possibly absorb it all. But there is no need to do so, because the burden is on the proponent of the new technique to show a scientific consensus supporting its use; if a fair overview of the literature discloses that scientists significant either in number or expertise publicly oppose that use of hypnosis as unreliable, the court may safely conclude there is no such consensus at the present time.

(1c) That is the case before us. On the topic of hypnotically aided recall we have reviewed numerous scientific treatises and articles in scholarly journals. From this review it clearly appears that major voices in the scientific community oppose the use of hypnosis to restore the memory of potential witnesses, with or without procedural safeguards, on the ground of its intrinsic unreliability. This unreliability is due both to certain properties of human memory and to factors inherent in the nature of hypnosis. We begin with the former, which have been little mentioned in the cases.

C

The professional literature, however, rejects this belief: the scientists who work in the field generally agree that, as Dr. Schafer testified at trial, the memory does not act like a videotape recorder, but rather is subject to numerous influences that continuously alter its content. This view has been expressed at least since the pioneer study of Sir Frederic C. Bartlett of Cambridge University, published a half-century ago. (Bartlett, Remembering (1932, reprinted 1964).) Using a different simile in that pretelevision era, Bartlett critically examined the conventional belief of his time that "traces" of every event were laid down in the mind and permanently stored until they were "re-excited" by a stimulus and reappeared as memories. Bartlett began his analysis by pointing out that "there are obvious difficulties. The traces are generally supposed to be of individual and specific events. Hence, every normal individual must carry about with him an incalculable number of individual traces. Since these are all stored in a single organism, they are in fact bound to be related one to another, and this gives to recall its inevitably associative character" ( id. at p. 203).

Less obvious but even more important was Bartlett's now-famous conclusion, drawn from his experimental work, that memory is productive rather than reproductive: "The first notion to get rid of is that memory is primarily or literally reduplicative, or reproductive. . . . In fact, if we consider evidence rather than presupposition, remembering appears to be far more decisively an affair of construction rather than one of mere reproduction." ( Id. at pp. 204-205.) As had often been shown, "condensation, elaboration and invention are common features of ordinary remembering" ( id. at p. 205). Expanding on the latter point, Bartlett reported that "our studies have shown us that all manner of changes in detail constantly occur in instances which every normal person would admit to be genuine instances of remembering. There are changes in order of sequence, changes of direction, of complexity of structure, of significance, . . ." ( Id. at p. 312.) In summary, "Remembering is not the re-excitation of innumerable fixed, lifeless and fragmentary traces. It is an imaginative reconstruction, or construction, built out of the relation of our attitude towards a whole active mass of organized past reactions or experience. . . ." ( Id. at p. 213.)

Bartlett's insight that "the past is being continually re-made, reconstructed in the interests of the present" ( id. at p. 309) prevailed in due course among his colleagues, and is now the generally accepted view of the profession. Of the many works that support this view, we shall discuss but two as examples. In his analysis of the "permanent memory" hypothesis, Professor Douglas L. Hintzman observes that "Before even considering the relevant data, we can see a number of reasons for thinking that true forgetting does occur. Throughout our adult lives brain cells die and are not replaced. Even those that survive undergo continual change. There is a constant 'turnover' of the chemicals that make them up, just as there is in other parts of the body. And as we deal with our changing environment, it seems likely that we encode many new experiences by modifying trace structures originally developed to deal with old ones." (Hintzman, The Psychology of Learning and Memory (1978) p. 298.) He then examines the two phenomena most often cited as "evidence" for the permanent memory hypothesis.

The first is the work of Wilder Penfield, a neurosurgeon who has performed numerous brain operations on patients with advanced cases of epilepsy. In the course of those operations Penfield discovered that when he stimulated various locations in the brain with an electric probe, the patients — who were conscious during the procedure — reported experiencing vivid and detailed "flashbacks" reminding them of events that assertedly occurred in their childhood. To explain these reports, Penfield proposed a model of memory very similar to the "videotape recorder" theory espoused by the police psychologists. Professor Hintzman points out several serious objections to Penfield's conclusions. First, on some occasions the patients report that they experience not a "memory" but a dream or a hallucination, or simply a feeling of deja vu. ( Id. at p. 301.) It is difficult if not impossible to distinguish between these phenomena. Next, "There is never any independent verification of the reported 'memories' — nothing to indicate the experience is really an event from the patient's past. There is only the subject's statement that it seems familiar, and therefore must be something that happened at an earlier time." ( Id. at pp. 301-302.) As we shall see, this lack of independent verification infects most of the claimed "evidence" of reliable hypnotic recall. Finally, "such reports have not been obtained from nonepileptic patients" ( id. at p. 302), and hence form a scientifically inadequate basis for drawing conclusions about the memory processes of the large majority of the population.

The second phenomenon often cited as evidence for permanent memory is "hypnotic age regression." This is the procedure by which, under hypnotic suggestion, a subject appears to regress in mental age until an earlier date in his life, then seems to "relive" the events he experienced on that date and their accompanying emotions. Again Professor Hintzman finds little persuasive value in such demonstrations, stressing that the subject's claim to recall specific individuals or events from his childhood is rarely if ever corroborated because of obvious difficulties in doing so. He also notes studies in which hypnotized subjects have instead undergone age "progression" — i.e., have been made to believe they are living 10 or more years in the future — and have reported their future "memories" with equal conviction and verisimilitude. "Good hypnotic subjects," the author explains, "will go to great lengths to comply with the hypnotist's requests, and this apparently includes constructing realistic scenarios and acting them out." ( Id. at p. 303.) Professor Hintzman concludes that while the permanent memory hypothesis is tantalizing in its simplicity, the evidence offered in its support is weak and the hypothesis is probably incorrect. ( Id. at p. 304.)

Elizabeth F. Loftus, Ph.D., a highly experienced investigator in the field of memory, reaches similar conclusions in a number of her articles (e.g., Loftus Loftus, On the Permanence of Stored Information in the Human Brain (1980) 35 Am. Psychologist 409) and in her valuable treatise, Eyewitness Testimony (1979). Adopting a different simile, she explains in the latter that "During the time between an event and a witness' recollection of that event — a period often called the 'retention interval' — the bits and pieces of information that were acquired through perception do not passively reside in memory waiting to be pulled out like fish from water. Rather, they are subject to numerous influences. External information provided from the outside can intrude into the witness' memory, as can his own thoughts, and both can cause dramatic changes in his recollection." ( Id. at pp. 86-87.) Her reasons for this view deserve close attention. On the basis of her research Dr. Loftus identifies a number of the influences that can cause a memory to change during the retention interval without the witness' awareness: the witness may "compromise" the memory with a subsequently learned but inconsistent fact ( id. at p. 56); he may "incorporate" into the memory a nonexistent object or event casually mentioned by a third party, e.g., in later questioning ( id. at p. 60); postevent information may change the way the witness "feels" about the original incident, e.g., may affect his impression of how noisy or how violent it was ( id. at pp. 70-72); because a witness is under great social pressure to be complete and accurate he may fill gaps in his memory by guessing, and thereafter "recall" those guesses as part of the memory; and if the witness is subjected to repeated questioning, any erroneous statement he made early on may be "frozen into" the memory and reappear later as a fact ( id. at pp. 84-86). There is no way to tell, moreover, whether any given detail recalled by the witness comes from his original perception or from external information that he subsequently acquired. ( Id. at p. 78.)

From these studies Dr. Loftus has no doubt that postevent experiences can alter any witness' memory, subtly but irreversibly. The author then considers and rejects the permanent memory hypothesis. Observing that "People cling to highly suspicious evidence to support this belief" ( id. at p. 115), she is particularly critical of the "evidence" allegedly provided by hypnotic age regression: "many investigators believe that hypnosis is unreliable and unpredictable, and is just as likely to create new memories as to recover old ones." ( Ibid.) And because such accounts are ordinarily unverifiable, "Vivid memories may be produced, but who can say whether these have or have not been altered by subsequent experiences to which a person has been exposed." ( Ibid.) Dr. Loftus concedes it will never be possible to "prove" conclusively that a witness does not have an unaltered memory trace of a given event, for it can always be objected that the technique for unearthing it was inadequate — i.e., that "we did not dig deep enough." ( Id. at p. 117.) Nevertheless, after careful investigation the author reports that "My colleagues and I have used a number of different techniques to try to induce such witnesses to reveal evidence of any traces of the original information. In all of these cases, we have been unable to provide any evidence that an intact original memory remains." (Italics added; id. at p. 118.)

The "videotape recorder" theory of law enforcement hypnotists also lacks empirical support for the third of its assumptions, to wit, that upon being unlocked by hypnosis the witness' repressed memories are "replayed" without further modification as he recalls the original event. Once more the research results are otherwise: in this final stage of the process, known as "retrieval," the accuracy of the witness' memory may be adversely affected by outside factors even as he recalls it. Again Dr. Loftus identifies some of those influences as follows: the witness may subconsciously tailor his recall to conform to expectations implied by the person questioning him; those expectations may be conveyed, intentionally or not, either by such conduct of the questioner as tone of voice, emphasis, pauses, facial expression and other "body language" ( id. at pp. 72-74), or by the particular method of interview used or precise form of question asked ( id. at pp. 90-94); and the witness may be more likely to respond to such cues if the questioner is a status figure (e.g., a doctor or a law enforcement official) than if he is merely a passerby inquiring what happened ( id. at pp. 97-98).

Lastly, Dr. Loftus warns there is no clear correlation between the witness' confidence in the accuracy of his recall and its accuracy in fact: indeed, studies have shown that in some circumstances "people can be more confident about their wrong answers than their right ones. To be cautious, one should not take high confidence as any absolute guarantee of anything." ( Id. at p. 101.) The final distorting influence on memory retrieval, then, is a well-documented phenomenon: "Most people, including eyewitnesses, are motivated by a desire to be correct, to be observant, and to avoid looking foolish. People want to give an answer, to be helpful, and many will do this at the risk of being incorrect. People want to see crime solved and justice done, and this desire may motivate them to volunteer more than is warranted by their meager memory. The line between valid retrieval and unconscious fabrication is easily crossed." ( Id. at p. 109.)

D

We turn, then, to the professional literature on the latter topic. For present purposes we need not add to this already lengthy discussion by analyzing that literature in detail; it will be enough if we simply set forth its principal relevant conclusions, with citations to a representative sample of supporting studies. The conclusions will necessarily be oversimplified, but full explanations of each point can be found in the cited authorities and similar works.

1. Hypnosis is by its nature a process of suggestion, and one of its primary effects is that the person hypnotized becomes extremely receptive to suggestions that he perceives as emanating from the hypnotist. The effect is intensified by another characteristic of the hypnotic state, to wit, that the attention of the subject is wholly focused on and directed by the hypnotist. The suggestions may take the form of explicit requests or predictions by the hypnotist; or they may be inferred by the subject from information he acquired prior to or during the hypnotic session, or from such cues as the known purpose of that session, the form of questions asked or comments made by the hypnotist, or the hypnotist's demeanor and other nonverbal conduct. The suggestions can be entirely unintended — indeed, unperceived — by the hypnotist himself.

2. The person under hypnosis experiences a compelling desire to please the hypnotist by reacting positively to these suggestions, and hence to produce the particular responses he believes are expected of him. Because of this compulsion, when asked to recall an event either while in "age regression" or under direct suggestion of heightened memory ("hypermnesia"), he is unwilling to admit that he cannot do so or that his recollection is uncertain or incomplete. Instead, he will produce a "memory" of the event that may be compounded of (1) relevant actual facts, (2) irrelevant actual facts taken from an unrelated prior experience of the subject, (3) fantasized material ("confabulations") unconsciously invented to fill gaps in the story, and (4) conscious lies — all formulated in as realistic a fashion as he can. The likelihood of such self-deception is increased by another effect of hypnosis, i.e., that it significantly impairs the subject's critical judgment and causes him to give credence to memories so vague and fragmentary that he would not have relied on them before being hypnotized.

3. During the hypnotic session, neither the subject nor the hypnotist can distinguish between true memories and pseudomemories of various kinds in the reported recall; and when the subject repeats that recall in the waking state (e.g., in a trial), neither an expert witness nor a lay observer (e.g., the judge or jury) can make a similar distinction. In each instance, if the claimed memory is not or cannot be verified by wholly independent means, no one can reliably tell whether it is an accurate recollection or mere confabulation. Because of the foregoing pressures on the subject to present the hypnotist with a logically complete and satisfying memory of the prior event, neither the detail, coherence, nor plausibility of the resulting recall is any guarantee of its veracity.

4. Nor is such guarantee furnished by the confidence with which the memory is initially reported or subsequently related: a witness who is uncertain of his recollections before being hypnotized will become convinced by that process that the story he told under hypnosis is true and correct in every respect. This effect is enhanced by two techniques commonly used by lay hypnotists: before being hypnotized the subject is told (or believes) that hypnosis will help him to "remember very clearly everything that happened" in the prior event, and/or during the trance he is given the sugggestion that after he awakes he will "be able to remember" that event equally clearly and comprehensively. Further enhancement of this effect often occurs when, after he returns to the waking state, the subject remembers the content of his new "memory" but forgets its source, i.e., forgets that he acquired it during the hypnotic session ("posthypnotic source amnesia"); this phenomenon can arise spontaneously from the subject's expectations as to the nature and effects of hypnosis, or can be unwittingly suggested by the hypnotist's instructions. Finally, the effect not only persists, but the witness' conviction of the absolute truth of his hypnotically induced recollection grows stronger each time he is asked to repeat the story; by the time of trial, the resulting "memory" may be so fixed in his mind that traditional legal techniques such as cross-examination may be largely ineffective to expose its unreliability.

IV

Second, when it is the defendant himself — not merely a defense witness — who submits to pretrial hypnosis, the experience will not render his testimony inadmissible if he elects to take the stand. In that case, the rule we adopt herein is subject to a necessary exception to avoid impairing the fundamental right of an accused to testify in his own behalf. ( People v. Robles (1970) 2 Cal.3d 205, 214-215 [ 85 Cal.Rptr. 166, 466 P.2d 710].)

Third, like the court in Mack (fn. 28, ante) we do not undertake to foreclose the continued use of hypnosis by the police for purely investigative purposes. We recognize that on occasions in the past a subject has apparently been helped by hypnosis to remember a verifiable fact — such as a license plate number — that the police previously did not know and were then able to use as a "lead" for further investigation of the crime. It is neither appropriate nor necessary for us to enter the debate as to the need for this investigative technique, or its reliability. We reiterate, however, that for the reasons stated above any person who has been hypnotized for investigative purposes will not be allowed to testify as a witness to the events that were the subject of the hypnotic session. Evidence discovered by such an investigation, of course, is not ipso facto rendered inadmissible by the prior hypnosis.

Fourth, error in admitting the testimony of a previously hypnotized witness is not reversible per se; its effect must still be judged under the prejudicial error test adopted in People v. Watson (1956) 46 Cal.2d 818, 836 [ 299 P.2d 243]. (See People v. Kelly, supra, at p. 40 of 17 Cal.3d.) The test is to be applied, however, in light of the reasons for our holding herein.

(1d) Arguing that the error was harmless, the Attorney General assumes that the inadmissible evidence in this record is limited to those few portions of Catherine's testimony that she asserted were directly affected by her hypnotic experience. (See pt. I B, ante.) Yet as we have seen, it is the consensus of informed scientific opinion today that in no case can a person previously hypnotized to improve his recollection reliably determine whether any unverified item of his testimony originates in his own memory or is instead a confusion or confabulation induced by the hypnotic experience. It would fly in the face of that consensus to allow a witness to be the judge of which portions of his testimony were actually produced by hypnosis.

The Attorney General suggests that we can at least determine which of Catherine's recollections were potentially the product of hypnosis, by the device of comparing her testimony at trial with the prehypnotic versions of her story given in her interview at the police station and her testimony at the preliminary hearing. But Dr. Schafer's testimony and the professional literature agree that the effects of pretrial hypnosis to restore a witness' recollection go beyond the bare production of pseudomemories during the trance: the experience will tend to clothe the witness' entire testimony in an artificial but impenetrable aura of certainty, and may distort the witness' recall of related events occurring both before and after the hypnotic session. Moreover, it would be impossible in most cases for an appellate court to undertake the kind of comparative analysis proposed by the Attorney General, because such materials as station-house interviews by the police or preliminary hearing testimony are not ordinarily part of the record.

We conclude that proper application of the Watson prejudicial error test in the present context requires the appellate court to determine whether it is reasonably probable that a result more favorable to the defendant would have occurred if the testimony of the previously hypnotized witness as to all matters relating to the events of the crime had not been admitted. This was the analysis we followed in Kelly. Applying the same analysis to the record before us, we find the error in admitting Catherine's testimony at trial to be prejudicial as it constituted virtually the sole incriminating evidence against defendant. To prevent a miscarriage of justice, a conviction predicated on such tainted evidence cannot be allowed to stand. (Cal. Const., art. VI, § 13.)

V

(6) Of defendant's remaining contentions, we need address only one that bears on the question of retrial. At the close of the prosecution's case-in-chief, defendant unsuccessfully moved for a judgment of acquittal on the ground of insufficiency of the evidence. (Pen. Code, § 1118.1.) He now contends the trial court erred in denying that motion, arguing that Catherine was incompetent as a witness because her intoxication had impaired her ability to perceive and remember the events of the evening, and that her testimony was so inconsistent as to be unbelievable. The effect of her intoxication, however, was for the jury to determine; on this record it falls far short of incompetence as a matter of law. And although her testimony was vague and self-contradictory on a number of points, when taken as a whole it was not inherently incredible and would have constituted at least "substantial evidence" to support a verdict of guilt. (See People v. Blair (1979) supra, 25 Cal.3d 640, 666; People v. Pierce (1979) 24 Cal.3d 199, 210 [ 155 Cal.Rptr. 657, 595 P.2d 91].)

It is true that we now hold Catherine's testimony legally inadmissible because of her pretrial hypnotic experience. But in the circumstances of this case the holding does not justify a judgment of acquittal. The purpose of a motion under section 1118.1 is to weed out as soon as possible those few instances in which the prosecution fails to make even a prima facie case. ( People v. Belton (1979) 23 Cal.3d 516, 520-521 [ 153 Cal.Rptr. 195, 591 P.2d 485].) It therefore speaks to "the evidence then before the court" (§ 1118.1), i.e., to the evidence that the trial court has properly admitted as of the time the motion is determined. As noted above (fn. 33, ante), none of the hypnosis cases we now follow had been decided as of the time of the motion herein, and hence the trial court applied the then-prevailing general rule that the fact of Catherine's pretrial hypnotic experience went to "the weight, not the admissibility" of her testimony.

(7) For the same reason, retrial is not prohibited by the federal double jeopardy clause under the rule of Burks v. United States (1978) 437 U.S. 1 [57 L.Ed.2d 1, 98 S.Ct. 2141], and Greene v. Massey (1978) 437 U.S. 19 [57 L.Ed.2d 15, 98 S.Ct. 2151], followed in this state ( People v. Pierce, supra, 24 Cal.3d at pp. 209-210). That rule forbids retrial after a reversal ordered because the evidence introduced at trial was insufficient to support the verdict. (See, e.g., In re Johnny G. (1979) 25 Cal.3d 543, 548-549 [ 159 Cal.Rptr. 180, 601 P.2d 196].) It is inapplicable, however, to the situation here presented. The rule achieves its aim — i.e., of protecting the defendant against the harassment and risks of unnecessary repeated trials on the same charge — by the device of giving the prosecution a powerful incentive to make the best case it can at its first opportunity. ( Burks, at p. 11 [57 L.Ed.2d at pp. 9-10].) But the incentive serves no purpose when, as here, the prosecution did make such a case under the law as it then stood; having done so, the prosecution had little or no reason to produce other evidence of guilt. To be sure, we now hold it error to admit Catherine's testimony against defendant; but "reversal for trial error, as distinguished from evidentiary insufficiency, does not constitute a decision to the effect that the government has failed to prove its case." ( Id. at p. 15 [57 L.Ed.2d at p. 12].) Rather, the matter is governed by the settled rule that the double jeopardy clause does not prohibit retrial after a reversal premised on error of law. ( Ibid.; accord, United States v. Tateo (1964) 377 U.S. 463, 465 [12 L.Ed.2d 448, 450, 84 S.Ct. 1587], and cases cited.)

It follows that there is no legal bar to retrying defendant on these charges. Of course, for the reasons stated above Catherine cannot be allowed to testify in such a trial on any of the events that were the subject of her hypnotic experience; her prehypnotic testimony at the preliminary hearing, however, may be admissible in lieu thereof. Whether a retrial is justified in the circumstances of this case is for the prosecutor to determine.

The judgment is reversed.

Bird, C.J., Newman, J., Broussard, J., and Tobriner, J., concurred.