Opinion
MOSK, J.In People v. Shirley (1982) 31 Cal.3d 18, 66-67 [181 Cal.Rptr. 243, 641 P.2d 775], we held that the use of hypnosis to restore or improve the memory of a potential witness is not accepted as a reliable procedure by a consensus of the relevant scientific community, and hence the testimony of such a witness is inadmissible as to all matters that were the subject of the hypnotic session. In the present case we address the question, left open in Shirley (id. at p. 67, fin. 53), whether the rule of that decision applies to witnesses who were hypnotized before its date (Mar. 11, 1982). As will appear, we conclude that the question must be answered in the affirmative—i.e., that Shirley applies to all cases not yet final as of the date it was decided. This conclusion is dictated by settled retroactivity precedents of this court and the United States Supreme Court, and is supported by the consistent practice of our sister states on the point. Applying the Shirley rule to the case at bar, we further conclude that the introduction of such hypnotically induced testimony was prejudicial error on the record before us, and hence compels reversal of the judgments of conviction.
I.
In count I of the information defendants Guerra and Murkidjanian were jointly charged with a single count of forcible rape; in counts II and III each *391was individually charged with a single count of attempted forcible oral copulation; and as to each count it was further alleged that defendants “voluntarily acted in concert” to commit the offense charged. (Pen. Code, §§ 264.1, 288a, subd. (d).) After two days of testimony the court declared a mistrial as to Murkidjanian because of the introduction of inadmissible and prejudicial hearsay evidence against him. The trial continued as to Guerra only, and the jury convicted him on both counts charged against him and found true the allegation of acting in concert. The court released Guerra on bail. A new trial of the charges against Murkidjanian was then held, and the jury convicted him on both counts and found true the allegation of acting in concert. The court denied motions for new trial by each defendant, sentenced Guerra to the lower term of five years’ imprisonment on count I, sentenced Murkidjanian to the middle term of seven years’ imprisonment on the same count, and stayed sentence on the other counts pending appeal and service of the sentences on count I, at which time the stay would become permanent. Defendants filed simultaneous notices of appeal, and their appeals have been consolidated.
A.
As in Shirley, the record of the case at bar discloses sharply conflicting stories.1 Judy S. testified she was 20 years old and worked as a supermarket cashier while attending state college. On February 25, 1979, she finished her shift at the market at midnight, watched a television program at a fellow worker’s house, then returned at 2 a.m. to the apartment she shared with a roommate. Shortly thereafter she went downstairs to an “open house” party being held in another apartment in the same building. There were 30 or 35 people at the party, including Murkidjanian and Guerra; quantities of alcohol were being drunk, and marijuana was being smoked. Judy mingled with the crowd and met a number of guests, but talked mainly to one Tom Logue. After about half an hour, Logue accompanied her back to her apartment and was invited in. Judy sat on the floor and worked on a school poster that she was making. Logue subsequently left for a few moments, instructing her not to lock the door. He reappeared with a beer and a marijuana cigarette, which she declined to share. Ten minutes later he went back to the party a second time, explaining that he would be riding home with two people and needed to see if they were still there. Judy soon heard another knock on her door; she said “Come in,” and Murkidjanian and Guerra entered. They said they were looking for Logue in order to give him a ride home; she remembered meeting them at the party, and asked them to *392have a seat. Guerra sat on the floor and watched her work, while Murkidjanian wandered around the apartment. Guerra left briefly to get a tape from his car. Judy began playing it on her stereo; he turned the volume up, and resumed his seat on the floor. Logue then entered the apartment for a third time, used Judy’s bathroom, chatted with everyone present, and departed again.
At this point the stories of the three main participants diverge. Judy testified that after Logue left, Murkidjanian grabbed her under the arms, lifted her to her feet, and started to carry her into her roommate’s bedroom. She protested vigorously and screamed for help; he put his hand over her mouth and she bit his thumb. When Murkidjanian began taking her sweater off, she asked if she could go to her bedroom and get a diaphragm so that she would not become pregnant. Holding her by the arm, Murkidjanian walked her back through the living room. Guerra, who had remained seated on the floor during these events, got to his feet and followed them into Judy’s bedroom. She testified that she “believed” Guerra closed and locked the door behind them, because Murkidjanian was occupied in holding onto her; she also testified that Guerra helped Murkidjanian take off her upper clothing. Thereafter, Guerra simply leaned against the dresser and watched. While Murkidjanian was undressing himself, Judy removed the rest of her clothes and inserted the diaphragm. Murkidjanian lifted her onto her bed, but she sat up against the headboard and kept her legs tightly closed. Murkidjanian straddled her on his hands and knees and sought to force her thighs apart with his elbows.
Judy testified, however, that Murkidjanian was unable to achieve an erection either then or at any time during the ensuing events. First he tried to reach an erection by masturbating, but was unsuccessful. Next he tried to have Judy orally copulate him for the same purpose: he “scooted up” on the bed and told her to “make him hard, and he wasn’t going to leave until [she] did.” She continued to resist vigorously, however, and he managed to get his penis only as far as her neck before giving up and resuming his attempt to masturbate.
On the question of penetration, Judy testified that Murkidjanian next inserted his finger into her vagina. The prosecutor then asked, “Did you feel his penis in your vagina?” and Judy answered, “Yes.” When told to describe how it felt, she replied, “round and soft.” And when asked specifically whether his penis was hard or flaccid inside her vagina, she answered, “Flaccid.” He never ejaculated.
According to Judy, during the encounter Guerra sat on a rocking chair and laughed at Murkidjanian’s discomfiture. At one point Judy saw Guerra *393apparently unzipping his pants, and she asked if he were next; he nodded and then said, “No, I’ll pass.” Thereafter Guerra came over to the bed, and she took his hand and squeezed it. Judy testified that Guerra told Murkidjanian, “Can’t you see the pain she’s in? Why don’t we just leave?” After a few minutes Murkidjanian gave up and exited from the room. Judy wrapped herself in a blanket; she testified that Guerra locked the door behind Murkidjanian and tried to calm her. He also explained to her, “if he had tried to help pull Murkidjanian off that there would just have been a big fight and that I did the best thing I could.” Judy told him she wanted to call her boyfriend, and he offered to dial the number for her. Guerra left while she spoke with her boyfriend; the latter took her to her mother’s house, and six or seven hours later she called the police to report the events.
On cross-examination Judy reiterated that throughout the whole episode Murkidjanian never managed to achieve an erection, and that except when he attempted to make her orally copulate him “he spent the rest of the time trying to get an erection” by self-manipulation. She also admitted that while Guerra was comforting her after Murkidjanian left, he hugged her and she hugged him back.2
Guerra’s testimony differed from Judy’s in several material respects. He was 21 years old, lived with his family and worked for his father. On the evening of February 24, 1979, he and his friend Murkidjanian met some young men at a drive-in, including Logue, and were invited to the party in Judy’s building. While there, they consumed two six-packs of beer between them. At one point in the evening Logue told them “there was some action going on upstairs. If we wanted any part of it.” They went up to Judy’s apartment, knocked, and entered when she said “Come in.” She offered them a seat, and Guerra “plopped” down on the floor and lay on his side. He testified that he left briefly to get a tape, but denied he turned up the volume on the stereo. He described Logue’s frequent visits to the apartment, and said that he learned from one of those visits that there was a bathroom in Judy’s room. He testified that he “wasn’t really paying any attention” when Murkidjanian lifted Judy up from the floor because he was listening to his tape, and he denied hearing her call for help. Shortly after Murkidjanian and Judy went into her bedroom, he entered in order to use the bathroom. He denied helping Murkidjanian take any clothes off Judy or deliberately closing the door; rather, he testified that the bathroom was located behind the bedroom door, and it was necessary to push the latter door partly closed in order to get to the bathroom. After spending a few minutes in the bathroom he came out and saw Murkidjanian and Judy lying *394naked on the bed; he confirmed that Murkidjanian did not have an erection and was attempting to masturbate.
Guerra further testified that as he emerged from the bathroom he was in the process of tucking in his shirt and zipping up his pants. Judy asked him if he were next and he replied, “No, thanks. I’ll pass.” At her request he took her hand and held it; Murkidjanian, meanwhile “was still down there trying to get himself erected.” Guerra testified that when Murkidjanian asked her to orally copulate him, Judy became very upset, “like she had wakened up from a bad dream or something.” Finally, Guerra confirmed that he urged Murkidjanian to stop and to leave the apartment, that he subsequently comforted Judy and explained why he had not intervened forcibly, and that they hugged each other as he departed.
Murkidjanian’s defense was consent. He testified that he was 21 years old and owned his own business, a catering truck. He corroborated Guerra’s testimony as to how they were invited to the party, what Logue told them about Judy, and the initial events at her apartment. He stated, however, that he had consumed two six-packs of beer himself and was drunk, and that he, too, used a bathroom in Judy’s apartment shortly after they arrived. Believing Judy wanted to “have some fun,” Murkidjanian testified, “I just sort of playfully scooped her off the floor.” He denied that she resisted, and said his belief in her cooperation was confirmed when she volunteered to get her diaphragm. He testified that he “helped” her take off her clothes and got on the bed with her. Throughout the remainder of the episode, however, he was “Trying to get an erection to make love with her. I could never get hard. I was too drunk, too drunk.” After failing in his attempt to masturbate, he asked Judy if she would orally copulate him for the same reason. She refused, and he became frustrated and angry. He testified that she suddenly began to cry, and this scared him into giving up the effort. He also confirmed that during these events Guerra had entered to use the bathroom, and thereafter held Judy’s hand and urged him to leave her alone. Finally, he denied that he ever achieved penetration.
B.
We turn to the evidence of hypnosis in this case, particularly with respect to the crucial issue of penetration.
Fullerton Police Officer Granados testified that he interviewed Judy on the morning after the assault. He was a graduate of the police academy, had attended rape seminars, and had conducted 15 to 20 preliminary rape investigations; from this experience he had learned the elements of the crime of rape. He specifically asked Judy if there had been any penetration, and *395she replied there had not.3 With respect to Guerra, Officer Granados further testified that Judy did not state that Guerra had helped undress her; on the contrary, she told the officer that Guerra “at no time had anything to do with the actual attempt[ed] rape . . . ,”4
A second investigating officer, Fullerton Detective Suffern, testified that he interviewed Judy two days later and again asked her if Murkidjanian had achieved penetration, and she again answered “No.” He nevertheless talked with Judy several more times in the ensuing two weeks, and finally asked her to come to the juvenile hall to be hypnotized on March 12, 1979. According to Judy, she knew before the hypnotic session that his purpose was to see if she could “remember” that she had been penetrated.5
The hypnotist was Fullerton Police Officer Davinroy. His training in hypnosis consisted of attending a four-day course given by the Law Enforcement Hypnosis Institute in January 1978, and a three-day course given by the same organization a year later.6 Before this case he had testified on the subject of hypnosis only once, in juvenile court. Officer Davinroy stated that prior to the hypnotic session with Judy, Detective Suffern told him “what he needed to know.” When defense counsel asked what that was, Davinroy replied that Suffern told him three people were suspected of an assault on Judy7 and he wanted “specific information as to what each of these people had done.” Suffern told him that Judy was having difficulty in remembering everything because “there was a block there that she needed help with.” When asked what else Suffern wanted, Officer Davinroy testified “there was some conversation about her not recalling penetration
*396When she appeared for the hypnotic session, Judy recognized Officer Davinroy: he had lived in her apartment building, and a few months earlier he had tried to fix her car for her. She had expressed gratitude for his efforts, and they thus began the session on a friendly basis. Fullerton Detective O’Brien was present throughout the session, and assisted by operating a tape recorder. According to Officer Davinroy, however, they did not succeed in recording the whole proceeding because Detective O’Brien “kept going into hypnosis” himself.8 In addition, Officer Davinroy talked with Judy for a few minutes before she was hypnotized and the recorder was turned on, and for a few minutes after she was brought out of the trance and the recorder was turned off.
Officer Davinroy testified that in his opinion Judy entered a hypnotic trance.9 While in the trance she related the events of the night of February 25, 1979, but did not mention any penetration. Officer Davinroy then began the process of bringing her to a lighter trance level as a preparatory step to dehypnotizing her. Among other things, he told her that when she came out of hypnosis she would be able to remember as much or as little of the events “as you’re comfortable with.” At that point Judy said, “I feel Greg [Murkidjanian] inside me ... a little . . . and it’s very soft.” Officer Davinroy asked her “what part” she felt, but her answers were equivocal. He then told her that the information she remembers after hypnosis, “you’ll remember quite vividly and accurately for as long as seems important.” He next implanted the specific posthypnotic suggestion that “if you recall more information later, which very often happens, I’d just like you to call Detective Suffern.”
Officer Davinroy testified that during their unrecorded discussion after Judy had been dehypnotized, she indicated to him that she “now remem*397bered” there had been sexual penetration. At the preliminary examination Judy positively testified that penetration had been achieved.
Prior to trial both defendants moved to suppress Judy’s testimony on the ground that hypnotically induced evidence is so unreliable that it should be excluded from criminal trials in this state. The prosecutor responded that “concededly, the only element of this entire crime she did not recall to the police officers was the penetration. The penetration is something that she accidentally recalled after she was hypnotized.” The prosecutor then argued that counsel’s sole remedy was to attempt to impeach the witness by pointing out the inconsistency. The court denied the motion, apparently on the theory that the fact of hypnosis goes to credibility rather than admissibility. At the outset of trial defendants again objected to Judy’s proposed testimony and moved to exclude it, contending that it was the product of hypnosis rather than of Judy’s memory of the events. After a brief hearing the court again refused to suppress the testimony.
At trial Judy admitted that the first time she told the police there had been penetration was during hypnosis. When asked by the prosecutor whether she had also recalled it before hypnosis, she replied, “Oh, I just kind of blocked it out of my mind.” And when asked whether she could now recall it because she had been hypnotized or because she remembered it from the events of the night of February 25, she asserted, “I can remember it from that night.”
The defense called Dr. Cameron Donald Pennock to testify as an expert on hypnosis.10 Dr. Pennock is a physician and psychiatrist who has had 30 years of training and practice in hypnosis; he has used hypnosis both for purposes of therapy and as a factfinding procedure; he has taught it to psychiatric resident physicians, and has written papers on the subject. Dr. Pennock described the phenomenon of hypnosis, and emphasized the dangers of unconscious suggestions and confabulation when the hypnotist is an authority figure like a police officer and is not an experienced professional in the medical or behavioral sciences. After listening to the tape recording of Officer Davinroy’s hypnotic session with Judy and reading a transcript of the recording, Dr. Pennock stated that in his opinion the hypnotic session was a case of “psychological brainwashing.” He explained that Officer Davinroy may not have known all the details of the crimes, but “he certainly was aware of the thing he was trying to prove”—even if unconsciously— and that was penetration; indeed, it was “the very thing that both of them [i.e., Officer Davinroy and Judy] knew that they were after.” Dr. Pennock testified he had discerned numerous instances in the tape and the transcript *398in which Officer Davinroy expressed to Judy his prehypnotic expectations of what the session would produce, as well as giving her subtle clues while she was in the trance that probably influenced her answers.11 Dr. Pennock concluded that in his opinion more than 50 percent of what Judy said under hypnosis was probably true, but that the remainder—including the claim of penetration—was false.
At the hearing on the motion for new trial the defense called Dr. Albert J. Rosenstein to testify as an expert witness.12 Dr. Rosenstein is a clinical psychologist who has studied hypnosis for 30 years and used it extensively in his practice. He confirmed Dr. Pennock’s views as to the inherent risk of eliciting false or inaccurate information under hypnosis because of the subject’s hypersuggestibility and eagerness to please the hypnotist. After hearing the tape recording of Officer Davinroy’s hypnotic session with Judy, and reading the police report of Officer Granados (fn. 3, ante) and Officer Davinroy’s report of the hypnotic session (fn. 9, ante), Dr. Rosenstein stated that in his opinion “the patient [i.e., Judy] responded in a programmed manner as though the patient had been led through the process of arriving at the conclusion that she was penetrated.” He did not find “any indication that the patient spontaneously, objectively, without being pushed, without being carried to the point, arrived at a conclusion that led me to believe that she was penetrated. ” In particular, Dr. Rosenstein testified that if a patient knows that one of the reasons she is being asked to undergo hypnosis is to determine whether she was sexually penetrated, that knowledge alone constitutes a suggestion that she “recall” such penetration. And if that “recall” occurs shortly after hypnosis, Dr. Rosenstein would be “very, very suspicious of it.”
II.
Both defendants contend that the case is controlled by the rule of People v. Shirley (1982) supra, 31 Cal.3d 18, and that if Shirley is applied to this record the admission of Judy’s posthypnotic testimony must be deemed both erroneous and prejudicial. Respondent does not take issue with the latter point, but vigorously urges that our rule barring posthypnotic testimony should not apply to any witness hypnotized before the date of the Shirley decision. We begin with that contention.
*399A.
To determine whether a decision should be given retroactive effect, the California courts first undertake a threshold inquiry: does the decision establish a new rule of law? If it does, the new rule may or may not be retroactive, as we discuss below; but if it does not, “no question of retroactivity arises,” because there is no material change in the law. (Donaldson v. Superior Court (1983) 35 Cal.3d 24, 36 [196 Cal.Rptr. 704, 672 P.2d 110] [plur. opn.]; People v. Garcia (1984) 36 Cal.3d 539, 547-548 [205 Cal.Rptr. 265, 684 P.2d 826]; United States v. Johnson (1982) 457 U.S. 537, 549 [73 L.Ed.2d 202, 213-214, 102 S.Ct. 2579].) In that event the decision simply becomes part of the body of case law of this state, and under ordinary principles of stare decisis applies in all cases not yet final. “As a rule, judicial decisions apply ‘retroactively.’ [Citation.] Indeed, a legal system based on precedent has a built-in presumption of retroactivity.” (Solem v. Stumes (1984) 465 U.S. 638 [79 L.Ed.2d 579, 586, 104 S.Ct. 1338, 1341].)
The most common examples of decisions that do not establish a new rule of law in this sense are those which explain or refine the holding of a prior case, those which apply an existing precedent to a different fact situation, even if the result may be said to “extend” the precedent, or those which draw a conclusion that was clearly implied in or anticipated by previous opinions (e.g., People v. Superior Court (Kiefer) (1970) 3 Cal.3d 807 [91 Cal.Rptr. 729, 478 P.2d 449], explained in Gallik v. Superior Court (1971) 5 Cal.3d 855, 859-860 [97 Cal.Rptr. 693, 489 P.2d 573]; see also id. at p. 860, fn. 4 [retroactivity of Mozzetti v. Superior Court (1971) 4 Cal.3d 699 (94 Cal.Rptr. 412, 484 P.2d 84)]).13
If the decision establishes a new rule of law, a second question arises: was there a prior rule to the contrary? If there was, the new rule— again—may or may not be retroactive, as we discuss below; if there was not, the new rule applies in all cases not yet final. This is so for the obvious reason that there cannot have been any justifiable reliance on an old rule when no old rule existed. And the emphasized word is crucial: “Unjustified ‘reliance’ is no bar to retroactivity.” (Solem v. Stumes (1984) supra, 465 U.S. 638 [79 L.Ed.2d 579, 589, 104 S.Ct. 1338, 1343].) It follows that *400“In all such cases the ordinary assumption of retrospective operation [citations] takes full effect.” (Donaldson, supra, 35 Cal.3d at p. 37.)
The latter principle is well settled: “As a matter of normal judicial operation, even a non-retroactive decision [i.e., one that cannot serve as a basis for collateral attack on a final judgment] ordinarily governs all cases still pending on direct review when the decision is rendered.” (People v. Rollins (1967) 65 Cal.2d 681, 685, fn. 3 [56 Cal.Rptr. 293, 423 P.2d 221].) For example, in People v. Aranda (1965) 63 Cal.2d 518, 528-529 [47 Cal.Rptr. 353, 407 P.2d 265], we adopted a new rule barring the admission into evidence of any portion of a codefendant’s extrajudicial statement that implicates the defendant; and in People v. Charles (1967) 66 Cal.2d 330, 334 [57 Cal.Rptr. 745, 425 P.2d 545], we held that “cases still pending on direct review should be adjudicated in accord with the principles which Aranda established. In reaching this conclusion, we adhere to the settled practice both of this court and of the United States Supreme Court.” We observed (ibid.) that “The historic pattern of applying the court’s current expression of a basic principle to cases pending on appeal finds numerous classic illustrations,” including People v. Kitchens (1956) 46 Cal.2d 260, 262 [294 P.2d 17] (applying Cahan to cases not yet final), and Linkletter v. Walker (1965) 381 U.S. 618, 622, footnote 4 [14 L.Ed.2d 601, 604, 85 S.Ct. 1731] (Mapp applies to cases not yet final). Referring to Rollins, we concluded (at p. 335) that “Earlier this year, we reaffirmed the principle implicit in all of these decisions and concluded that convictions should ordinarily be tested on appeal under the law then applicable, not the law prevailing at the time of trial. [Citation.] We see no reason to depart today from this basic postulate of appellate review.” Nor have we departed from it since. (See, e.g., In re Jimenez (1978) 21 Cal.3d 595, 608-609 [147 Cal.Rptr. 172, 580 P.2d 672] (citing Charles); People v. Gainer (1977) 19 Cal.3d 835, 853 [139 Cal.Rptr. 861, 566 P.2d 997] (same).)
Examples of decisions that establish a new rule when there was no prior rule to the contrary are noted as follows in Donaldson (35 Cal.3d at p. 37): “Neither is there any issue of retroactivity when we resolve a conflict between lower court decisions, or address an issue not previously presented to the courts.”14 The latter category, of course, also includes instances in which the issue was presented but not squarely decided in the prior opinions, or in which prior Court of Appeal decisions resolving it were vacated by grants of hearing in this court (People v. Garcia (1984) supra, 36 Cal.3d at p. 549). In each of these cases there was no clear rule on which anyone could have justifiably relied.
*401This second class of decisions may be further defined by reference to its opposite, i.e., the decisions that establish a new rule when there was a prior contrary rule. According to both Johnson (457 U.S. at p. 551 [73 L.Ed.2d at pp. 214, 215]) and Donaldson (35 Cal.3d at p. 37), such a “clear break with the past” occurs only in certain limited situations, i.e., when the decision (1) explicitly overrules a precedent of this court (e.g., People v. Disbrow (1976) 16 Cal.3d 101, 113 & fn. 14 [127 Cal.Rptr. 360, 545 P.2d 272]), or (2) disapproves a practice impliedly sanctioned by prior decisions of this court (e.g., Donaldson, supra, 35 Cal.3d at p. 37), or (3) disapproves a longstanding and widespread practice expressly approved by a near-unanimous body of lower-court authorities (e.g., People v. Bustamante (1981) 30 Cal.3d 88, 96, 102 [177 Cal.Rptr. 576, 634 P.2d 927]).
It is only in this “narrow class of decisions” (Johnson, supra, 457 U.S. at p. 553 [73 L.Ed.2d at p. 216]) that there can have been justifiable reliance on an old rule to the contrary, and hence that the courts may choose to make, on grounds of policy, an exception to “the ordinary assumption of retrospective operation” (Donaldson, supra, 35 Cal.3d at p. 37).15 At the present time the California courts decide whether to make such an exception by weighing the three factors summarized in Stovall v. Denno (1967) 388 U.S. 293, 297 [18 L.Ed.2d 1199, 1203]: “(a) the purpose to be served by the new standards, (b) the extent of the reliance by law enforcement authorities on the old standards, and (c) the effect on the administration of justice of a retroactive application of the new standards.” (E.g., Donaldson, at p. 38 of 35 Cal.3d; In re Joe R. (1980) 27 Cal.3d 496, 511-512 [165 Cal.Rptr. 837, 612 P.2d 927]; People v. Kaanehe (1977) 19 Cal.3d 1, 10 [136 Cal.Rptr. 409, 559 P.2d 1028].)
The traditional description of the Stovall test as “tripartite,” however, is misleading. To begin with, the second and third factors it identifies are really but two sides of the same coin: when the retroactive application of a new rule causes a substantial effect on the administration of justice, it is primarily because there was a substantial reliance on the old rule by law enforcement authorities; conversely, when that reliance was minimal, retroactive application will usually have a similarly minimal effect on the administration of justice. The second and third factors are thus largely duplicative; functionally, the test is bipartite.
Even that description is inadequate in the many cases in which the first factor—the purpose of the new rule—points plainly towards retroactivity or *402prospectivity. The determination of that purpose is then both the beginning and the end of the inquiry: “the factors of reliance and burden on the administration of justice are of significant relevance only when the question of retroactivity is a close one after the purpose of the new rule is considered.” (In re Johnson (1970) 3 Cal.3d 404, 410 [90 Cal.Rptr. 569, 475 P.2d 841]; accord, Desist v. United States (1969) 394 U.S. 244, 251 [22 L.Ed.2d 248, 256, 89 S.Ct. 1030].) When that purpose clearly favors retroactivity or prospectivity, it will be given effect without regard to the weight of the remaining factors. (Ibid.) In all such cases the test is essentially unipartite.
B.
Perhaps the most consistent application of this principle has been in cases in which the primary purpose of the new rule is to promote reliable determinations of guilt or innocence. The United States Supreme Court has aptly characterized its operation in that event: “Where the major purpose of new constitutional doctrine is to overcome an aspect of the criminal trial that substantially impairs its truth-finding function and so raises serious questions about the accuracy of guilty verdicts in past trials, the new rule has been given complete retroactive effect. Neither good-faith reliance by state or federal authorities on prior constitutional law or accepted practice, nor severe impact on the administration of justice has sufficed to require prospective application in these circumstances. ” (Fn. omitted.) (Williams v. United States (1971) 401 U.S. 646, 653 [28 L.Ed.2d 388, 395, 91 S.Ct. 1148].) Numerous decisions of the high federal court so holding prior to Williams are summarized in our opinion in Johnson, supra, 3 Cal.3d at p. 411.16
Since Williams, the high court has held a new rule retroactive on this ground in three more instances. In Ivan V. v. City of New York (1972) 407 U.S. 203 [32 L.Ed.2d 659, 92 S.Ct. 1951], the issue was the retroactivity of the rule of In re Winship (1970) 397 U.S. 358 [25 L.Ed.2d 368, 90 S.Ct. 1068], which required the states to comply with the standard of proof beyond a reasonable doubt in juvenile proceedings. The court observed that the Winship rule was “a prime instrument for reducing the risk of convictions resting on factual error” (id. at p. 363 [25 L.Ed.2d at p. 375]); and quoting the foregoing language of Williams, the court concluded that “the major purpose of the constitutional standard of proof beyond a reasonable doubt announced in Winship was to overcome an aspect of a criminal trial *403that substantially impairs the truth-finding function, and Winship is thus to be given complete retroactive effect.” (407 U.S. at p. 205 [32 L.Ed.2d at pp. 661-662].)
In Hankerson v. North Carolina (1977) 432 U.S. 233 [53 L.Ed.2d 306, 97 S.Ct. 2339], the issue was the retroactivity of the rule of Mullaney v. Wilbur (1975) 421 U.S. 684 [44 L.Ed.2d 508, 95 S.Ct. 1881], which barred the states from shifting to the defendant the burden of persuasion on any element of the crime. The court held the rule retroactive under Ivan V., reasoning that “In Mullaney, as in Winship, the rule was designed to diminish the probability that an innocent person would be convicted and thus to overcome an aspect of a criminal trial that ‘substantially impairs the truth-finding function.’ ” (Id. at p. 242 [53 L.Ed.2d at p. 315].) The state sought to distinguish Ivan V. by stressing the factors of reliance and impact on the administration of justice.17 Because of the truth-finding purpose of the Mullaney rule, however, the high court declined to give these factors any weight. The court explained (at pp. 243-244 [53 L.Ed.2d at pp. 315-316]) that “It is true that we have said that the question of whether the purpose of a new constitutional rule is to enhance the integrity of the factfinding process is a question of ‘degree,’ [citation]; and when the degree to which the rule enhances the integrity of the factfinding process is sufficiently small, we have looked to questions of reliance by the State on the old rule and the impact of the new rule on the administration of justice in deciding whether the new rule is to be applied retroactively. [Citations.] But we have never deviated from the rule stated in Ivan V. that ‘“[w]here the major purpose of new constitutional doctrine is to overcome an aspect of the criminal trial that substantially impairs its truth-finding function and so raises serious questions about the accuracy of guilty verdicts in past trials, the new rule [is] given complete retroactive effect.”’ [Citation.] The reasonable-doubt standard of proof is as ‘substantial’ a requirement under Mullaney as it was in Winship. ” (Fn. omitted.)
Finally, in Brown v. Louisiana (1980) 447 U.S. 323 [65 L.Ed.2d 159, 100 S.Ct. 2214], the high court gave retroactive effect to the rule of Burch v. Louisiana (1979) 441 U.S. 130 [60 L.Ed.2d 96, 99 S.Ct. 1623], which required unanimity for conviction of a nonpetty offense by a six-person jury. The court reviewed the various formulations of the doctrine here in issue, and found their common denominator to be that “any rule which raises substantial doubts about the reliability of the jury’s verdict should be applied *404retroactively.” (447 U.S. at p. 330, fn. 6 [65 L.Ed.2d at p. 167] [plur. opn.].) The opinion then concluded that the Burch rule was designed to preserve the right to trial by jury and assure the reliability of its verdict, and hence that its purpose “clearly requires retroactive application.” (Id. at p. 334 [65 L.Ed.2d at p. 169].)
During the past two years the United States Supreme Court has twice more grappled with retroactivity questions, in United States v. Johnson (1982) supra, 457 U.S. 537, and in Solem v. Stumes (1984) supra, 465 U. S. 638 [104 S.Ct. 1338]. Neither opinion, regrettably, has succeeded in providing the definitive answers to all such questions; perhaps because they represent the viewpoints of different majorities, the opinions do not even appear to be wholly consistent. For our present purposes, however, we need not attempt to reconcile these two efforts;18 it is enough to note that both opinions agree on the continued vitality of the doctrine of cases like Ivan V. Thus Johnson purports to limit its rule of retroactivity to Fourth Amendment decisions, but immediately adds the significant observation that “The logic of our ruling, however, is not inconsistent with our precedents giving complete retroactive effect to constitutional rules whose purpose is to overcome an aspect of the criminal trial that substantially impairs its truth-finding function,” citing Hankerson and Ivan V. (457 U.S. at p. 562, fn. 21 [73 L.Ed.2d at p. 222]; see also id. at pp. 544, 548-549 & fn. 11 [73 L.Ed.2d at pp. 209-210, 212-214].) In turn, Solem reiterates that “Complete retroactive effect is most appropriate where a new constitutional principle is designed to enhance the accuracy of criminal trials,” citing Williams. (465 U.S. 638 [79 L.Ed.2d at p. 587, 104 S.Ct. at p. 1342].) And the opinion holds that the rule there in issue is not “the sort of decision that goes to the heart of the truthfinding function, which we have consistently held to be retroactive,” citing Brown, Hankerson, and Arsenault. (Id. at p. 645 [79 L.Ed.2d at p. 588, 104 S.Ct. at p. 1343].)
C.
The decisions of this court are in full accord with the foregoing federal doctrine. Thus in In re Montgomery (1970) 2 Cal.3d 863 [87 Cal.Rptr. 695, 471 P.2d 15], we gave full retroactive effect to the rule of Barber v. Page (1968) 390 U.S. 719 [20 L.Ed.2d 255, 88 S.Ct. 1318], i.e., that the de*405fendant is denied his constitutional right of confrontation and cross-examination when he is convicted on the transcript of the preliminary hearing testimony of an absent witness, unless the prosecution has made a good-faith effort to secure the witness’s attendance at the trial. We reasoned that even if the impact of that rule may be significant, the error presents “a serious risk that the issue of guilt or innocence may not have been reliably determined.” (2 Cal.3d at p. 866.) We observed that “The United States Supreme Court has consistently accorded full retroactivity to rules of criminal procedure fashioned to correct serious flaws in the fact-finding process at trial. [Citations.] The denial of the right to confrontation and cross-examination is such a serious flaw.” {Id. at p. 867.)
In In re Johnson (1970) supra, 3 Cal.3d 404, 413, after a thorough review of federal precedents, we concluded that “the more directly the new rule in question serves to preclude the conviction of innocent persons, the more likely it is that the rule will be afforded retrospective application. Further, if the rule relates to characteristics of the judicial system which are essential to minimizing convictions of the innocent, it will apply retroactively regardless of the reliance of prosecutors on former law, and regardless of the burden which retroactivity will place upon the judicial system. ” (Italics added.) We explained that “The overwhelming concern of recent retroactivity decisions with the relation of the rule in question to the reliability of the truth-determining process at trial is but a corollary to the ultimate test of the integrity of the judicial process: its capacity to ensure the acquittal of the innocent.” {Id. at p. 416.)
In People v. Burnick (1975) 14 Cal.3d 306, 332 [121 Cal.Rptr. 488, 535 P.2d 352], we held that the federal and state due process clauses require the standard of proof beyond a reasonable doubt in proceedings leading to involuntary commitment as a mentally disordered sex offender, and “because the major purpose of this rule is to overcome an aspect of those proceedings which ‘substantially impairs the truth-finding function,’ our decision today must be given complete retroactive effect,” citing Ivan V. (Accord, People v. Thomas (1977) 19 Cal.3d 630, 644-645 [139 Cal.Rptr. 594, 566 P.2d 228] (holding that the requirements of proof beyond a reasonable doubt and jury unanimity in narcotics addicts commitment proceedings are fully retroactive under Burnick and Ivan V.).)
In People v. Gainer (1977) supra, 19 Cal.3d 835, we held it error to give the so-called “Allen instruction” to potentially deadlocked juries, primarily because it impaired the defendant’s right to the independent judgment of each juror and a truly unanimous verdict, and exerted undue pressure on the dissenting jurors to vote with the majority simply to reach a verdict of any kind. (Id. at pp. 848-851.) We further held this new rule to be retro*406active, explaining (at p. 853) that “it is aimed at judicial error which significantly infects the fact-finding process at trial. [Citation.] Given this critical purpose, neither judicial reliance on previous appellate endorsements of the charge in this state nor any effects on the administration of justice require us to deny the benefit of this rule to cases now pending on appeal.” We noted that this principle applies not only to new constitutional doctrine but also to new “judicially declared rules which are not necessarily constitutionally required.” (Id. at p. 853, fn. 18.)
Finally, in Pryor v. Municipal Court (1979) 25 Cal.3d 238 [158 Cal.Rptr. 330, 599 P.2d 636], we construed Penal Code section 647, subdivision (a), to criminalize as “lewd or dissolute conduct” only certain acts committed in a limited manner and with a specific intent. {Id. at p. 244.) We further declared this new rule to be retroactive, reasoning that the purpose of the holding “implicates questions of guilt and innocence, for conduct which a trier of fact might have found criminal under the older vague definition may clearly fall beyond the scope of the statute as construed in the present case.” {Id. at p. 258.) Relying on Gainer, we concluded {ibid.) that the rule would at least apply to all cases not yet final, regardless of the degree of reliance on the old rule and the burden placed on the administration of justice by such application. (See also People v. Garcia (1984) supra, 36 Cal.3d 539, 549.)19
III.
We apply the foregoing analysis to our holding in Shirley that testimony of a witness who has been hypnotized to restore or improve his memory is inadmissible as to all matters that were the subject of the hypnotic session.
A.
First, it is undisputed that Shirley established a new rule of law: it did more than merely explain a prior holding or apply an existing precedent to different facts, and it was not clearly foreshadowed by previous decisions of this court.
Second, upon careful review of the case law it appears there was no previous California rule to the contrary, within the meaning of the retroac*407tivity precedents. Few opinions of our courts had mentioned the phenomenon of hypnosis before we discussed it in depth in Shirley, and fewer still had addressed its reliability as a device to restore the memory of a witness. One of the earliest of these opinions set the tone for decades to come. In People v. Ebanks (1897) 117 Cal. 652, 665-666 [49 P. 1049], the defendant offered to prove he had been hypnotized by an expert hypnotist, and while under hypnosis had made statements on the basis of which the expert was prepared to testify that he was innocent. The trial court refused the offer on the sweeping ground that “ ‘The law of the United States does not recognize hypnotism,’” and this court upheld the ruling with the equally confident observation that “We shall not stop to argue the point and only add the court was right.” (Italics in original.)
Over 60 years passed before hypnosis was again at issue in a decision of this court. In Cornell v. Superior Court (1959) 52 Cal.2d 99, 102 [338 P.2d 447], the counsel for a defendant awaiting trial sought a writ of mandate to compel the sheriff to allow an expert hypnotist to examine his client in jail for the purpose of learning his whereabouts on the night of the crime. This court granted the writ on the ground that the defendant’s constitutional right to counsel includes the right to consult before trial with his attorney and the latter’s right to a reasonable opportunity to discover facts in order to prepare a defense. We recognized that Ebanks held inadmissible any statements a defendant makes while under hypnosis, but reasoned that such statements might nevertheless serve as leads to the discovery of other evidence that might itself be admissible and might constitute a defense to the charge.
In People v. Busch (1961) 56 Cal.2d 868, 878 [16 Cal.Rptr. 898, 366 P.2d 314], a defense physician offered to give his opinion of the defendant’s mental state at the time of the crimes, based on pretrial examinations of the defendant that had included hypnosis. The prosecution objected on the ground that hypnosis was not a sufficiently scientific means of exploring a person’s state of mind, and that the witness was not qualified. The trial court ruled that the witness was competent to testify as a medical doctor on the defendant’s mental condition, but not to testify insofar as his opinion was based on the results of hypnosis. This court affirmed the ruling on the ground that “the trial judge did not act unreasonably in his determination that a proper foundation was not established as to the reliability of an analytical tool still seeking recognition in the field of psychiatry, or as to the qualifications of this particular witness to give an opinion on the state of mind of the accused . . . ."
In People v. Modesto (1963) 59 Cal.2d 722, 733 [31 Cal.Rptr. 225, 382 P.2d 33], the trial court allowed a defense psychiatrist to give her opinion of the defendant’s mental state at the time of the crimes, based on pretrial *408examinations of the defendant that had included hypnosis, but the court excluded under Busch a tape recording of statements that defendant made while hypnotized. This court distinguished Busch on the ground that in the case before it the witness was qualified as an expert psychiatrist, and that nothing in Busch “would preclude introducing in evidence all of the data on which she based her opinion.” The court conceded, however, that the tape recording might properly have been excluded in the exercise of the trial judge’s discretion to weigh probative value against the risk of prejudice, and held it error not to have exercised that discretion.
Finally, in People v. Blair (1979) 25 Cal.3d 640, 665 [159 Cal.Rptr. 818, 602 P. 2d 738], we upheld a ruling excluding a tape recording of statements made by a defense witness while hypnotized and offered under the hearsay exception for past recollection recorded. We distinguished Modesto on the ground that such statements may be admissible as a basis for an expert opinion, and reaffirmed the Ebanks rule that they are otherwise inadmissible for the truth of the matters asserted “because the reliability of such statements is questionable.” Even though the witness was not the defendant but a disinterested observer, we explained that “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.)
It is obvious that in none of the foregoing decisions did this court hold, contrary to Shirley, that the testimony of a witness who has been hypnotized to restore his memory is admissible in the courts of California. Thus Shirley did not explicitly overrule a precedent of this court—the first category of decisions that can constitute a “clear break with the past” under Johnson and Donaldson.
Nor does it fall within the second category of such decisions, i.e., those which disapprove a practice impliedly sanctioned by prior decisions of this court. Respondent argues that none of the foregoing opinions “intimated” the Shirley rule; but that fact merely proves that the Shirley rule was new, not that our prior cases had impliedly approved a contrary rule. Rather, the repeated references in those opinions to the unreliability and inadmissibility of hypnotically induced testimony make it inconceivable that anyone could fairly have read them as encouraging the use of hypnosis to refresh a witness’s memory.
Nor, finally, does Shirley fall within the third category of “clear break” decisions, i.e., those which disapprove a longstanding and widespread prac*409tice expressly approved by a near-unanimous body of lower-court authority. To begin with, while we do not doubt that prior to Shirley there were a number of instances in which certain police departments in California used hypnosis in an attempt to restore or improve a witness’s memory, it does not appear that it was both a “longstanding” and “widespread” practice within the meaning of the retroactivity precedents. Contrary to the former routine destruction of the evidence of breathalyzer tests, for example, respondent fails to show that pretrial hypnosis of witnesses was “the standard and almost uniform procedure for law enforcement officials throughout the state” (People v. Hitch (1974) 12 Cal.3d 641, 654 [117 Cal.Rptr. 9, 527 P.2d 361]).
More important, there was no express approval of any such practice by a near-unanimous body of lower-court precedents. Respondent relies on only two cases. In the first, People v. Colligan (1979) 91 Cal.App.3d 846, 850 [154 Cal.Rptr. 389], a robbery victim was hypnotized by the police in an attempt to have her recall the license plate number of the getaway car, and while under hypnosis she also gave a description of the robber. Thereafter she identified the defendant at a photographic lineup, at the preliminary hearing, and at trial. On appeal the defendant complained that the latter identification was tainted by the pretrial hypnosis. Rejecting the contention, the Court of Appeal relied principally on the rule that “a claim of improper pretrial identification will not be considered on appeal absent an objection in the trial court . . . .” The court added that in any event “defendant does not contend that hypnotic suggestions were actually made to [the witness] which affected her identification”; and the court remarked in dictum, “We decline to hold that the use of hypnosis to help a witness remember a license number per se invalidates identification of a person seen and heard by that witness.”
In the second case, People v. Diggs (1980) 112 Cal.App.3d 522 [169 Cal.Rptr. 386], an assault victim was hypnotized by the police for memory-enhancement purposes after the first trial ended in a hung jury. Prior to retrial the defendant moved to suppress the victim’s testimony because it was the product of hypnosis. At the hearing Dr. Bernard Diamond testified for the defense that hypnotically induced testimony is inherently unreliable, creates false memories, and makes the witness resistant to cross-examination; his testimony was supported by affidavits of other experts. The prosecution psychiatrist, however, expressed the opinion that hypnotic memory enhancement is reliable, and described the procedure he used to hypnotize the victim in the case at hand. The trial court refused to suppress the testimony, and the defendant was convicted. The Court of Appeal reversed the judgment for lack of counsel at a postindictment lineup. {Id. at pp. 528-529.) For guidance in the event of a third trial in the case, the court also *410addressed the question of the admissibility of posthypnotic testimony, which it correctly described as “an issue not yet decided in California” (id. at p. 530). The court reasoned that the issue was governed by the Kelly-Frye rule (People v. Kelly (1976) 17 Cal.3d 24 [130 Cal.Rptr. 144, 549 P.2d 1240]; Frye v. United States (D.C.Cir. 1923) 293 F. 1013), i.e., that evidence based on a new scientific technique is admissible only on a showing that it is generally accepted as reliable in the scientific community in which it developed. The court then concluded (at p. 531) that “Dr. Diamond’s testimony to the contrary,” the testimony of the prosecution psychiatrist amounted to “sufficient evidence” of the reliability of hypnotic memory-enhancement to support the trial court’s ruling, apparently on the general theory (see Evid. Code, § 411) that the testimony of one witness is enough to prove any fact.20
The most that respondent is able to say of Colligan and Diggs is that they gave “implied support” for the use of hypnosis to restore a witness’s memory. But the retroactivity precedents require that any such practice be expressly approved by the lower courts before a contrary ruling of this court can be said to constitute a “clear break with the past.” And even if Diggs could be broadly read as approving hypnotic memory-enhancement in all cases rather than merely on the record before it, its lone voice did not amount to the required “near-unanimous body” of lower-court authority on the issue.21
Respondent also invokes a number of decisions of the lower federal courts and the courts of other states that had admitted hypnotically induced testimony prior to Shirley, either with or without certain “safeguards,” on the theory that the fact of hypnosis “goes to the weight, not the admissibility” of such evidence.22 We need not reach the question whether cases from other jurisdiction can determine if a decision of this court should be restricted to prospective effect in California, because the foreign case law on hypnosis did not constitute a “near-unanimous” body of precedents at the time *411of Shirley. While the cases relied on by respondent may have constituted the “prevailing rule” when the Shirley case went to trial (31 Cal.3d at p. 54, fn. 33), the law was nevertheless “in a state of flux” (id. at p. 32). In the four years preceding our Shirley decision there arose a contrary line of cases from still other states holding that hypnotically induced testimony was inadmissible under the Kelly-Frye test, and by the time we decided Shirley there were approximately as many jurisdictions rejecting such testimony as allowing it.23 Accordingly, even if we had the power to “resolve” in Shirley that conflict in foreign case law, the existence of the conflict ipso facto meant that our decision would still be governed by “the ordinary assumption of retrospective operation” (Donaldson, supra, at p. 37 of 35 Cal.3d).
We could end our analysis at this point: because there was no “old rule” to the contrary in California, Shirley did not constitute a “clear break with the past” and hence must be given normal application to all cases not yet final.
B.
Out of an excess of caution, however, we shall go further and assume arguendo that Shirley did amount to a “clear break” which in appropriate circumstances could be limited to prospective operation under the Stovall test. We proceed to apply that test.
As explained above (Part II, ante), the first step is to determine the major purpose of the new rule. On this point the Shirley opinion speaks for itself. After establishing that the admissibility of hypnotically aided recall is to be determined by the Kelly-Frye test, we reviewed numerous articles on hypnosis in scientific treatises and journals. “From this review,” we concluded, “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.” (31 Cal.3d at p. 56.) We then identified a number of reported causes of that unreliability.
First we noted studies of human memory showing that a witness often subconsciously alters his recollection of an event in response to later information or questioning, and does so in ways that are both irreversible and impossible to detect. (Id. at pp. 57-62.) Next we reviewed the scientific literature establishing that “each of the phenomena found by such research *412to contribute to the unreliability of normal memory reappears in a more extreme form when the witness is hypnotized for the purpose of improving his recollection.” (Id. at p. 63.) We summarized the salient conclusions of that literature as follows: A hypnotized person is hyperreceptive to suggestions by the hypnotist, whether express or implied, verbal or nonverbal, deliberate or unintended, even unperceived by the hypnotist himself. The hypnotized person experiences a compelling desire to please the hypnotist by giving the responses he believes are expected; his critical judgment is impaired, causing him to credit vague recollections he would not have relied on before hypnosis; and if he cannot actually recall an event he will produce a “pseudomemory” that may be compounded of irrelevant facts from an unrelated experience, “confabulations” or fantasized material unconsciously invented to fill the gaps, and conscious lies. If the claimed memory cannot be verified by independent means, neither the subject nor an expert nor a lay observer can distinguish between accurate recollections and the foregoing pseudomemories, either during the hypnotic session or afterwards, i.e., at the trial. Finally, a witness who is uncertain of his memory of an event before being hypnotized will become convinced by that process that the story he told under hypnosis is entirely true; that conviction grows stronger as he repeats the story, until “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.” (Id. at p. 66.) We concluded (ibid.) that the professional literature thus “demonstrates beyond any doubt that at the present time the use of hypnosis to restore the memory of a potential witness is not generally accepted as reliable by the relevant scientific community,” and hence that the testimony of a witness whose memory has been tampered with by hypnosis is inadmissible as to the events that were the subject of the hypnotic session.
We need not decide whether all decisions excluding evidence under the Kelly-Frye test promote the reliability of the factfinding process to such a degree that they should be given retroactive effect. The kind of evidence that the Shirley rule addresses is not simply a dry, technical explanation of an arcane method of scientific proof by a disinterested criminalist. (See, e.g., People v. Palmer (1978) 80 Cal.App.3d 239, 250-255 [145 Cal.Rptr. 466, 1 A.L.R.4th 1056] (analysis of gunshot residue particles by studying their chemical characteristics with a scanning electron microscope equipped for X-ray spectrometry).) Rather, it is often—as in Shirley and the case at bar—emotionally charged and highly personal testimony by the actual victim of or eyewitness to the crime being prosecuted, directly identifying the defendant as the perpetrator or furnishing some crucial piece of evidence against him. A rule that excludes such testimony when it is irremediably tainted by hypnosis is manifestly designed “ “‘to overcome an aspect of the *413criminal trial that substantially impairs its truth-finding function and so raises serious questions about the accuracy of guilty verdicts in past trials”’” (Hankerson v. North Carolina (1977) supra, 432 U.S. 233, 243 [53 L.Ed.2d 306, 316]). Under the federal and California precedents discussed earlier, the Shirley rule therefore applies to all cases not yet final. “Neither good-faith reliance” on prior law or accepted practice “nor severe impact on the administration of justice has sufficed to require prospective application in these circumstances.” (Williams v. United States (1971) supra, 401 U.S. 646, 653 [28 L.Ed.2d 388, 395].)
In a fruitless effort to distinguish the retroactivity precedents, respondent contends that hypnosis is different because its effects on the witness can assertedly be identified at trial and be counteracted by cross-examination. But both of these claims were expressly rejected by the consensus of the scientific community at the time of Shirley (31 Cal.3d at pp. 65-66, 68-69); and as will appear (Part IV, post), respondent fails to show that the consensus has significantly changed today. Respondent also argues that hypnosis is different because it presents “a mere potential for harm” and in any given case may have had no effect at all on the testimony in question. Yet the same can also be said, for example, of the standard of proof required in Burnick and the “dynamite instruction” condemned in Gainer: in any given case the result might well have been the same under the old rule. In all three instances the purpose of the new rule is to avoid a grave risk of tampering with the truth-finding process and convicting the innocent.
It follows that whether or not it replaced a prior rule to the contrary, the Shirley decision applies to all cases not yet final in our courts.24
C.
We find further support for our holding in the fact that in virtually every sister state in which hypnotically induced testimony has been declared inadmissible, the rule has also been applied in cases pending on appeal at the time it was adopted.
Thus in State v. Mena (1981) 128 Ariz. 226 [624 P.2d 1274], the Arizona Supreme Court held posthypnotic testimony inadmissible for all the reasons we thereafter invoked in Shirley (see 31 Cal.3d at pp. 46-48). The Mena court applied the equivalent of our Watson test to the error in admitting the posthypnotic testimony of the victim, and found prejudice because “had the *414victim’s testimony been excluded at trial” a different verdict would have been reasonably probable. (624 P.2d at p. 1280.) We applied the same test of prejudice, of course, in Shirley. (31 Cal.3d at p. 70.)
Mena was decided in February 1981. Eleven months later the Arizona Supreme Court reaffirmed the Mena rule in State ex rel. Collins v. Superior Court, etc. (1982) 132 Ariz. 180 [644 P.2d 1266]. Applying the Stovall test, the court purported to hold Mena “prospective only.” (Id. at p. 1276.) But the quoted phrase apparently meant simply that Mena could not be applied on collateral attack to cases that were final; as to cases still on direct appeal, the court declared a rule indistinguishable in effect from the rule we now adopt. After holding that any person hypnotized post-Mena (or hypnotized pre-Mena but who had not yet testified) was incompetent to be a witness, the court went on to explain: “We further hold that any conviction presently in the appeals process in which there was hypnotically recalled testimony and where the testimony and hypnosis took place pre-Mena will be examined on a case-by-case basis to determine if there was sufficient evidence, excluding the tainted testimony, to uphold the conviction. In other words, the question in these cases is whether the introduction of posthypnotic testimony was harmless error. ” {Ibid.; italics added.) Thus the court assumed that even on an appeal from a pre-Mena trial, hypnotically recalled testimony is “tainted” and its introduction was therefore “error”; the only issue is whether the error is prejudicial under a Watson-type standard. This is precisely what we hold today.25
Confirming the foregoing analysis, the Arizona Supreme Court subsequently applied its Mena rule to numerous cases pending on appeal at the time Mena was decided. Depending on the state of the evidence, in some cases the court held that the error in admitting posthypnotic testimony was harmless (e.g., State v. Thomas (1982) 133 Ariz. 533 [652 P.2d 1380, 1384-1385]), while in others it held the error prejudicial and reversed the judgment of conviction (e.g., State v. Stolp (1982) 133 Ariz. 213 [650 P.2d 1195, 1196] (aggravated assault)).
*415Another authority we relied on in Shirley (31 Cal.3d at pp. 42-44) was State v. Mack (Minn. 1980) 292 N.W.2d 764, a case in which the Minnesota Supreme Court held posthypnotic testimony inadmissible on the same grounds as Shirley. In the following year, the Minnesota court decided State v. Koehler (Minn. 1981) 312 N.W.2d 108, an appeal from a conviction of a bludgeon-murder of a 12-year-old girl. The only direct evidence linking the defendant to the crime scene was the testimony of a witness who identified the defendant’s car after he had been hypnotized by a police officer. The court noted that Mack was decided subsequently to the trial of the case {id. at p. 110, fn. 2), and a dissenting justice contended that “The trial court did not have the benefit of our decision in State v. Mack when this case was tried and the standards set forth in that case should not be applied retroactively.” (Id. at p. 110.) Nevertheless, the majority of the Minnesota Supreme Court held the admission of this testimony to be prejudicial error under the Mack rule. It therefore reversed the conviction with directions that on retrial the witness should not be permitted to testify about any matters adduced for the first time after hypnosis.26
The same sequence of rulings has also been followed in states in which the decision to exclude hypnotically recalled testimony came after we decided Shirley. Thus in Com. v. Kater (1983) 388 Mass. 519 [447 N.E.2d 1190], the Massachusetts Supreme Court held, for all the reasons stated in Shirley and its predecessors, that it was error to admit testimony by a previously hypnotized witness as to any matters remembered after the hypnotic session, and that error would be reversible if there was a substantial likelihood that a miscarriage of justice had occurred. Less than four months later the same court applied that rule to a case pending on appeal at the time Kater was decided. (Com. v. Brouillet (1983) 389 Mass. 605 [451 N.Ed.2d 128].) The court stated that it did not need to discuss the admissibility of posthypnotic testimony “because the case is controlled entirely” by Kater. The victim had not identified the defendant as her assailant until after the hypnotic session. The court held simply that “Kater requires the exclusion” of such testimony, and the judge should have granted the defendant’s pretrial motion to suppress it. {Id. at p. 130.) The court was well aware that it was applying Kater retroactively, noting “The judge did not, of course, have the benefit of Commonwealth v. Kater, supra, at the time this case was tried. Thus, his action was not wholly unjustified. Indeed, given the *416law at that time, he followed the correct procedure in holding a hearing” on the suppression motion. (Ibid., fn. 5.) Nevertheless the court held that the admission of the posthypnotic testimony was both error and prejudicial, and reversed a conviction of aggravated rape. (Accord, Com. v. Dodge (1984) 391 Mass. 636 [462 N.E.2d 1363].)
In State v. Collins (1983) 296 Md. 670 [464 A.2d 1028], the highest court of Maryland for the first time held posthypnotic testimony inadmissible under the Frye test.27 On the same day, the same court applied Collins to four pending appeals; it reversed the conviction in two (State v. Metscher (1983) 297 Md. 368 [464 A.2d 1052] (forcible oral copulation and assault with intent to maim); Grimes v. State (1983) 297 Md. 1 [464 A.2d 1065] (assault with intent to murder)), affirmed in a third (Simkus v. State (1983) 296 Md. 718 [464 A.2d 1055]), and in the fourth it remanded to the intermediate appellate court for reconsideration in light of Collins (State v. McCoy (1983) 297 Md. 5 [464 A.2d 1067]).
In People v. Gonzales (1982) 415 Mich 615 [329 N.W.2d 743], modified at 417 Mich. 968 [336 N.W.2d 751], the Supreme Court of Michigan likewise held that hypnotically induced testimony was inadmissible on Shirley grounds. Among the cases pending on appeal at the time Gonzales was decided was People v. Nixon (1982) 114 Mich.App. 233 [318 N.W.2d 655, 657-658], in which the Michigan Court of Appeals had affirmed a conviction of two counts of first degree murder despite the use of posthypnotic testimony of an eyewitness. The Michigan Supreme Court ordered the Nixon case remanded to the court of appeals “for reconsideration in light of People v. Gonzales . . . .” (330 N.W.2d 855, 856.) On remand, the court of appeals ruled that admission of the posthypnotic testimony was prejudicial error under Gonzales, and reversed the judgment. (People v. Nixon (1983) 125 Mich.App. 807 [337 N.W.2d 33].)
In People v. Hughes (1983) 59 N.Y.2d 523 [466 N.Y.S.2d 255, 453 N.E.2d 484], the New York Court of Appeals also held, for the reasons we set forth in Shirley, that hypnotically induced testimony is inadmissible under the Frye test. In People v. Perrino (1983) 96 App.Div.2d 952 [466 N.Y.S.2d 408], the trial court admitted posthypnotic testimony of two prosecution witnesses on the theory that the fact of hypnosis affects credibility only, and the defendant was convicted of aggravated assault. While the case was on appeal Hughes was decided; the intermediate appellate court applied the Hughes rule, and reversed the conviction.
*417Finally, in State v. Peoples (1984) 311 N.C. 515 [319 S.E.2d 177], the North Carolina Supreme Court overruled its decision to the contrary in State v. McQueen (1978) 295 N.C. 96 [244 S.E.2d 414], held hypnotically refreshed testimony inadmissible for the reasons discussed in Shirley, and declared the new rule applicable to all cases pending on appeal (319 S.E.2d at pp. 188-189).
Respondent suggests no reason why we should not follow this consistent practice of our sister states.
IV.
In the alternative, respondent attacks the Shirley rule itself. He contends that subsequent developments in the scientific literature and the case law have undermined the foundations of that decision to such an extent that it should be “reconsidered,” i.e., overruled. As will appear, however, those foundations have only been strengthened by recent scientific studies and persuasive decisions from our sister states. Like Mark Twain’s comment on his obituary, any report of the death of Shirley is an exaggeration.
A.
Respondent relies on two articles by professionals in the behavioral sciences.28 The first is by David Spiegel, M.D., a psychiatrist experienced in the use of hypnosis for therapeutic purposes. He is also, however, a proponent of the forensic use of hypnosis, and the content of his article lives up to the polemical tone of its title—The Shirley Decision: The Cure is Worse than the Disease. The piece is not, for example, a dispassionate, objective report of a new clinical or experimental study of the reliability of posthypnotic testimony that might show some change in the consensus of the scientific community identified in Shirley. Rather, it is a frontal assault on the Shirley decision itself—and through it, on Dr. Bernard Diamond, one of the principal participants in that consensus. The major portion of the article sets forth a series of disagreements between Drs. Spiegel and Diamond on such matters as how significant is the risk of intentional or unintentional cueing by the hypnotist, how often the hypnotized subject produces confabulations or other pseudomemories, how difficult it is to distinguish true from false memories, how resistant the subject will be to cross-exam*418ination at trial, and how effective are procedural precautions in minimizing the dangers of using hypnosis to restore a witness’s memory.
It is not our function, however, to resolve any such disputes within the scientific community, any more than it is our task to determine whether posthypnotic testimony is reliable as a matter of scientific fact. As we explained in Shirley (31 Cal.3d at p. 56), the courts view the professional literature on hypnosis “as ‘evidence,’ not of the actual reliability of the new scientific technique, but of its acceptance vel non in the scientific community.” That is all the Frye test requires, and all we undertook in Shirley. The appearance of Dr. Spiegel’s critique does not ipso facto destroy the consensus we identified in Shirley. The Frye test does not demand the impossible-proof of an absolute unanimity of views in the scientific community before a new technique will be deemed reliable; any such unanimity would be highly unusual, especially in the field of behavioral sciences. Rather, the test is met if use of the technique is supported by a clear majority of the members of that community. Even at the time of Shirley we were well aware there was a minority view on the risks of admitting posthypnotic testimony.29 Dr. Spiegel’s article is a vigorous restatement of that view; but as will appear, it remains nevertheless a minority.30
Still less persuasive is the second article relied on by respondent, Hypnotic Hypermnesia: A Critical Review, by Helmut Relinger, Ph.D. Dr. Re-linger is a psychologist who undertakes in this article simply to review and summarize a number of earlier studies by other experimenters on the topic of hypnotic hypermnesia, i.e., hypnotically improved recall by means of direct suggestion.31 According to Dr. Relinger, these studies suggest that *419hypnosis does enhance recall provided the material is meaningful and the subjects are allowed to respond by free narrative rather than by answering specific questions. The article is irrelevant, however, to the issue whether the consensus identified in Shirley is different today: all the studies that it discusses predate that decision, and we were aware of their conclusions when we addressed the matter in our opinion.32 Moreover, while the article reports on instances of increased recall under hypnosis, it does not focus on the equally crucial question whether such increased recall is accurate.
That question has been addressed in a number of articles published since Shirley, and therefore bearing on our inquiry. They are remarkably uniform in their conclusion that if there is any increased recall in hypnotic hypermnesia, it is purchased at the price of increased errors and probably also an increase in the subject’s misplaced confidence in those errors. Thus one study tested the effect of hypnosis on the recognition of faces; it found that hypnosis does not improve the ability to make a facial identification, but significantly increases the proportion of confident but mistaken identifications or “false alarms.” (Wagstaff, Hypnosis and Recognition of a Face (1982) 55 Perceptual & Motor Skills 816.) Another article reported on two studies using material typical to a police investigation. In the first, the results “[did] not support the hypothesis that hypnosis improves memory for items relevant to forensic investigation; if anything, the hypnotic treatment, possibly due to effects on arousal and motivation, had a somewhat deleterious effect.” (Wagstaff et al., Hypnosis and Eyewitness Memory: Two Experimental Analogues (1982) 10 IRCS Med. Sci.: Psychology & Psychiatry 894.) In the second study hypnosis again produced no improvement in memory but did result in a significantly greater proportion of false positive identifications. {Id. at p. 895.)33
*420A subsequent paper reported similar results from an elaborate study based on a series of mock assassination attempts. (Timm, A Theoretical and Empirical Examination of the Effects of Forensic Hypnosis on Eyewitness Recall, a paper presented at the 9th Internat.Cong. of Hypnosis & Psychosomatic Med., Glasgow, Scotland, Aug. 22-27, 1982.) It showed that hypnotized subjects tended to give fewer “I don’t know” answers to investigative questioning, that many of the statements they gave in lieu of that response were incorrect, and that the subjects were more confident after hypnosis, even of their erroneous answers. (Id. at pp. 36-37.) The author found it likely that these results reflected the demand characteristics of investigative hypnosis, and that “each time the subjects repeated their story it became harder for them to differentiate whether they were recalling their originally encoded memories or bits of information they acquired later.” (Id. at p. 37.) The author concluded that “it appears forensic hypnosis or other related procedures might prove helpful in certain situations, but only when sound questioning procedures are utilized and acquiring possible additional leads seems more important to solving the case than the necessity for that witness being as accurate and untainted as possible. ” (Italics added.) (Id. at pp. 40-41.)
Two recent studies confirm these conclusions. In the first, a hypnotic hypermnesia experiment (see fn. 31, ante) was designed to test not only the number of additional items recalled but also the accuracy of those memories. (Dywan & Bowers, The Use of Hypnosis to Enhance Recall (1983) 222 Sci. 184.) The outcome was clear: hypnosis produced a modest increase in the number of new items remembered, but at a considerable cost in accuracy: although the hypnotized subjects reported twice as many new items (both correct and incorrect) as the control subjects, they made three times as many new errors as the latter. (Id. at pp. 184-185.) The authors suggested that the increase in recall under hypnosis may not represent an actual improvement in ability to remember but rather a decrease in critical judgment, i.e., “less caution by subjects in what they are willing to report as memories.” (Id. at p. 185.) It may also be due to an increase in the vividness of the mental images generated as possible memories during recall attempts, and “the enhanced vividness could lead to a false sense of recognition and hence the inflated output as well as the surprising certainty that subjects have about their hypnotically enhanced recall.” (Ibid.) For all these reasons, the authors conclude that the results of their study “should give pause to those advocating the use of hypnosis in situations in which the veridicality [i.e., truth] of information is of prime concern.” (Ibid.)
Shortly thereafter a second article reported on an experiment designed to test whether a pseudomemory could be successfully implanted by hypnosis. (Laurence & Perry, Hypnotically Created Memory Among Highly Hypnotiz*421able Subjects (1983) 222 Sci. 523.)34 Each person in a group of 27 subjects selected a night during the previous week in which he or she had no memory of dreaming or awakening during the night. The subjects were then hypnotized, were instructed to relive the night in question, and were asked whether they heard some loud noises that awakened them (i.e., a suggested auditory hallucination). All but 10 said they could hear the noises, and they were encouraged to describe them. After the subjects were dehypnotized, fully half claimed they had actually heard the noises on the night in question. Indeed, they maintained that position even when told that the noises had in fact been suggested to them by the hypnotist, and they invented various rationales to justify their confidence in the “memory.” {Id. at p. 524.) In the authors’ view {ibid.), these results support Dr. Orne’s position that “the memories of victims and witnesses of crime can be modified unsuspectingly through the use of hypnosis. They suggest, further, that an initially unsure witness or victim can become highly credible in court after a hypnotic memory ‘refreshment’ procedure.”35 The authors concluded that in a real-life situation, where the subject is more emotionally involved and more motivated to cooperate, “Such ‘recall’ could lead to a false but positive identification and to all of the legal procedures and penalties that this implies.” {Id. at p. 584.)
As if to remind us that such real-life situations do occur, another article reported two actual cases in which the victim of a violent assault in the nighttime was unable to identify her assailant until after her memory had been “refreshed” by hypnosis, but in retrospect each identification was evidently the product of cues and assurances from the hypnotist, posthypnotic suggestions, and confabulations by the witness. (Karlin, Forensic Hypnosis—Two Case Reports (1983) 31 Internal.!. Clinical & Experimental Hypnosis 227.) The author, who has served as an expert for both prosecution and defense in hypnosis cases, acknowledged there is no way of knowing the extent to which these examples are typical; but they illustrate all too well his conclusion that “the great care which one must take in the forensic use of hypnosis must be redoubled when hypnosis is used to obtain an eyewitness identification of a perpetrator. It also suggests, to the degree that *422such anecdotes can, that excluding ‘hypnotically refreshed’ testimony may be the least of the evils available to our courts in this regard.” (Id. at p. 233.)36
A review of these and similar studies identified four principal effects of using hypnosis to restore memory that render it inherently unreliable. (Perry & Laurence, The Enhancement of Memory by Hypnosis in the Legal Investigative Situation (1983) 24 Can. Psychology 155.) First, the authors observed that the demand characteristics of the process cannot be eliminated: for example, when as frequently occurs the police ask to hypnotize a witness whom they have already questioned in some detail, the witness naturally infers there must be additional useful information “stored” in his memory that can be “retrieved” by the hypnosis. Second, the subject (and often the police hypnotist as well) shares a widespread but erroneous belief that hypnotized persons always tell the truth. Third, “the videotape” technique commonly used by law enforcement hypnotists gives the subject an apparently “scientific” rationale for believing the reality of every “memory” that may occur to him during hypnosis. Fourth, even if the hypnotist does not know all the facts of the crime under investigation, he will normally have a hypothesis about what happened in the case that it may be difficult for him to avoid communicating to the subject. The authors conclude that “Given all of these ingredients, and remembering that the victim or witness of crime is a person with a fragmented memory of an event, the investigative situation may be especially vulnerable to inadvertent cueing of the subject. Cueing becomes a very likely and damaging possibility given that, in hypnosis, the person is being asked, implicitly, to set aside critical judgement. To the extent that the person is able to do this, fantasy may reign supreme, even in a relatively unhypnotizable individual.” (Id. at p. 163.)
Finally, and most important, Dr. Martin T. Orne and three of his colleagues recently published a comprehensive evaluation of the current scientific data on hypnosis and its reliability as a memory-enhancing device. (Orne et al., Hypnotically Induced Testimony, in Eyewitness Testimony: Psychological Perspectives (Wells & Loftus edits. 1984) p. 171 [hereafter Hypnotically Induced Testimony].) The study confirms in detail each of the conclusions summarized in Shirley (31 Cal.3d at pp. 63-66), i.e., that a person under hypnosis is hyperreceptive to intended and unintended cues and eager to please the hypnotist; that hypnosis impairs the subject’s capacity to judge the reality of his memories; that he will produce on demand a recollection of an event which may be a compound of actual facts, irrelevant *423matter, and highly plausible “confabulations”; that without objective verification no one can distinguish true recollections from pseudomemories; and that hypnosis artificially increases the subject’s confidence in both his true and his false memories. (Orne et al., Hypnotically Induced Testimony, at pp. 173-178.) The authors also report a number of additional effects of hypnosis, not noted in Shirley, that contribute to the unreliability of posthypnotic testimony. Some of these effects increase the pressure on the hypnotized individual to produce pseudomemories.37 Other effects of hypnosis artificially enhance the subject’s credibility as a witness.38
After discussing Shirley and the decisions that preceded and followed it, the authors agree that “The present state of scientific knowledge is consistent with the rulings of a number of state supreme courts that memories retrieved through hypnosis are sufficiently unreliable that their use is precluded as eyewitness testimony in criminal trials. The nature of hypnosis and of its effects on memory leads to the possibility that beliefs of the hypnotist or subject may be transformed into inaccurate memories that the subject reports, believes, and subsequently is willing to testify to under oath. There is currently no available method for eliminating this possibility or for accurately determining in real-life situations the balance of increased recall versus increased distortion that may occur following hypnosis, because ground truth cannot be known with certainty.” {Id. at p. 204.)
Of particular interest is Dr. Orne’s repudiation of the approach taken in his name in the well-known decision of State v. Hurd (1981) 86 N.J. 525 [432 A.2d 86], In Hurd the New Jersey Supreme Court held that the *424posthypnotic testimony of an eyewitness would be admissible in criminal trials provided the hypnotic session was conducted in compliance with a set of procedural guidelines previously recommended by Dr. Orne himself. (Id. at pp. 96-97.) In the present article, however, Dr. Orne explains that “Although the recommended guidelines for conducting the hypnosis session help determine what was done during the session, they do not prevent (nor is there any reliable way to prevent) subjects from confounding distorted hypnotic memories with prior and subsequent nonhypnotic recall or from placing undue confidence in these distorted recollections.” (Italics added; Orne et al., Hypnotically Induced Testimony, supra, at p. 210.) He therefore believes that pretrial hypnosis should be allowed only for the limited purpose of seeking verifiable “leads” in a criminal investigation: “If the sole purpose of the hypnotic session is to provide clues that ultimately lead to the collection of independent evidence, its application becomes a means to an end that is no different from the use of other unreliable sources by the police.” (Italics added; id. at p. 196.)39 In all other cases—including “the situation in which a witness is hypnotized ostensibly only for investigative purposes but later testifies in court concerning recollections of the event in question” (ibid.)—the grave risks inherent in pretrial hypnosis outweigh its modest potential benefits. (Ibid.) Dr. Orne and his colleagues therefore conclude by warning once again that “There is no way, however, by which anyone (including an expert with extensive experience in hypnosis) can for any particular piece of information obtained in hypnosis determine whether it is an actual memory or a confabulation. For these reasons, hypnotically induced testimony is not reliable and ought not be permitted to form the basis of testimony in court.” (Id. at p. 211.)
Contrary to respondent’s claim, accordingly, it is evident in the light of the recent literature that the consensus of the scientific community on the reliability of hypnotically induced testimony remains unchanged since Shirley.
B.
We turn to respondent’s contention that the case law since Shirley has in effect repudiated our decision and now holds that hypnosis is a reliable technique for refreshing the recollection of witnesses. Again the claim seems to be composed of equal parts of exaggeration and wishful thinking.
*425It is true that the courts of several states have declined to follow Shirley and its predecessors on the general question of the admissibility of testimony recalled after hypnosis. But on close examination it appears the decisions relied on by respondent in fact represent no “new trend” at all: they sharply disagree among themselves as to their holdings, and they offer no rationale that has not already been considered and rejected by other cases on hypnosis. Thus the Washington court adopted the rule of State v. Hurd (N.J. 1981) supra, 432 A.2d 86, i.e., it agreed that posthypnotic testimony must comply with the Frye test, but held such testimony admissible provided there was strict compliance with the Orne “safeguards.” (State v. Long (1982) 32 Wn.App. 732 [649 P.2d 845],)40 Taking a different tack, Wisconsin rejected the Frye test and declared that the Hurd procedures are “merely guidelines, not imperatives,” yet followed Hurd in requiring a pretrial hearing on the “suggestiveness” of the hypnotic session in each case. (State v. Armstrong (1983) 110 Wis.2d 555 [329 N.W.2d 386, 391-395].) Florida adhered to the old rule of Harding v. State (1968) 5 Md.App. 230 [246 A.2d 302, 306], i.e., that the fact of hypnosis affects the credibility or weight of the testimony rather than its admissibility, but the court nevertheless “commended” the Hurd procedures to its trial courts. (Key v. State (Fla.App. 1983) 430 So.2d 909, 912.) North Dakota followed the Harding rule but flatly rejected the Hurd procedures. (State v. Brown (N.D. 1983) 337 N.W.2d 138, 146-152.)41 Idaho rejected all the foregoing tests and adopted its “own rule,” directing the trial court to determine in each case whether posthypnotic testimony is sufficiently reliable to merit admission in the “totality of the circumstances”—which includes, nevertheless, consideration of a “modified version of the Orne safeguards.” (State v. Iwakiri (1984) 106 Idaho 618 [682 P.2d 571, 578].) Finally, two states have simply held posthypnotic testimony admissible when it is “substantially the same” as the witness’s prehypnotic statements. (State v. Seager (Iowa 1983) 341 N.W.2d 420, 431; State v. Wren (La. 1983) 425 So.2d 756, 759; see also State v. Hutchison (1983) 99 N.M. 616 [661 P.2d 1315, 1319-1320].)
The Harding approach, of course, was rejected not only by Shirley and its predecessors but by the entire Hurd line of cases as well (see Shirley, 31 Cal.3d at pp. 36-39), and indeed has been repudiated by the very jurisdiction that adopted it in the first place. (Polk v. State (Md.App. 1981) supra, 427 A.2d 1041; State v. Collins (Md. 1983) supra, 464 A.2d 1028.) In turn, after careful consideration and for multiple reasons, we rejected the Hurd approach in Shirley (at pp. 39-40 of 31 Cal.3d), and numerous deci*426sions both before and after Shirley have joined us in doing so.42 We also rejected in Shirley, again on several grounds, the claim that posthypnotic testimony is harmless if it is “substantially the same” as prehypnotic statements. (Id. at p. 69.)
In contrast to this disarray of conflicting and discredited theories, the basic rule we adopted in Shirley—that hypnotically induced testimony is inadmissible per se—continues to draw new adherents.43 We have previously cited in another context (Part III C, ante) the recent decisions so holding by the distinguished high courts of New York (People v. Hughes (1983) supra, 453 N.E.2d 484, 494-495), Massachusetts (Com. v. Kater (1983) supra, 447 N.E.2d 1190, 1195-1199), Michigan (People v. Gonzales (1982) supra, 329 N.W.2d 743, 745-748), Maryland (State v. Collins (1983) supra, 464 A.2d 1028, 1032-1045), and North Carolina (State v. People (N.C. 1984) supra, 319 S.E.2d 177, 186-189). To those must now be added similar decisions by the courts of Colorado (People v. Quintanar (Colo.App. 1982) supra, 659 P.2d 710, 711-713), Indiana (Peterson v. State (1983) - Ind. — [448 N.E.2d 673, 675-678]), and Oklahoma (Robison v. State (Okla.Crim.App. 1984) 677 P.2d 1080, 1084-1085), and the first decision by a federal appellate court holding posthypnotic testimony “inadmissible whatever procedural safeguards were used to attempt to sanitize the hypnotic session” (United States v. Valdez (5th Cir. 1984) supra, 722 F.2d 1196, 1203).44
Certain of these decisions (e.g., Collins) quote extensively from Shirley, and all follow the spirit if not the letter of our decision on the principal issue at hand: they agree that the consensus of the scientific community continues to oppose the use of hypnosis to restore the memory of potential witnesses on the ground that it is inherently unreliable and impairs the de*427fendant’s right of confrontation, and all therefore hold hypnotically induced testimony inadmissible in their respective jurisdictions.
C.
Finally, respondent emphasizes that many of the foregoing courts have nevertheless declared a willingness to allow, under a variety of restrictions, the introduction of “prehypnotic evidence”; i.e., in certain circumstances the witness may be permitted to testify to facts that he remembered before he was hypnotized. (E.g., State ex rel. Collins v. Superior Court, etc. (Ariz. 1982) supra, 644 P.2d 1266, 1295-1296; Com. v. Kater (Mass. 1983) supra, 447 N.E.2d 1190, 1197-1199; State v. Patterson (1983) 213 Neb. 686 [331 N.W.2d 500, 501-504]; People v. Hughes (N.Y. 1983) supra, 453 N.E.2d 484, 495-497.) There is a superficially appealing simplicity to this view: because the evidence in question is recalled before the hypnotic experience, the argument goes, it must ipso facto be untainted by that experience. The rub, however, is that the testimony by which that evidence is put before the jury is given after the witness has been hypnotized. Unconcerned with any problems this sequence might create, respondent urges us to carve out an exception to the Shirley rule allowing such evidence to be introduced in California.
To begin with, we point out that the Shirley decision itself held admissible one type of prehypnotic evidence. The victim in that case testified in detail at the preliminary hearing before she was hypnotized. For the guidance of the court on remand we explained that although the victim would not be allowed to testify at retrial as to the events that were the subject of the hypnotic session, her prehypnotic testimony at the preliminary hearing would be admissible in lieu thereof, provided only that her disqualification had not been procured by the prosecution for the purpose of preventing her from testifying. (31 Cal.3d at pp. 71-72 & fn. 60.) We adhere to that rule. Preliminary hearing testimony is presented in open court, in the presence of the defendant, at a time when the defendant is represented by counsel, in a proceeding in which the defendant has both the motive and the opportunity to cross-examine the witness, and the testimony is reported and preserved in a formal court transcript. In these circumstances the introduction of preliminary hearing testimony at trial, when the witness is legally unavailable, does not violate either the hearsay rule (Evid. Code, § 240) or the defendant’s constitutional right of confrontation (see Ohio v. Roberts (1980) 448 U.S. 56 [65 L.Ed.2d 597, 100 S.Ct. 2531]; California v. Green (1970) 399 U.S. 149 [26 L.Ed.2d 489, 90 S.Ct. 1930]).
*428But the same cannot necessarily be said of other types of prehypnotic evidence.45 In Shirley we noted that a proposal to admit what is perhaps the most common form of such evidence—trial testimony by the witness as to facts he claims to have remembered before hypnosis—has been judicially criticized “for a number of scientific and practical reasons that we find persuasive. ” (Italics added.) (31 Cal.3d at p. 48, fn. 29, citing State ex rel. Collins v. Superior Court, etc. (Ariz. 1982) supra, 644 P.2d 1266, 1297-1299 (dis. opn. of Gordon, Vice C. J.).) We have reviewed the opinions relied on by respondent, but nothing therein convinces us to recant that finding. Indeed, one of those opinions itself criticizes the proposed exception for prehypnotic evidence as “difficult to reconcile” with the rule barring the testimony of a witness who has been subjected to pretrial hypnosis. (State v. Brown (N.D. 1983) supra, 337 N.W.2d 138, 149.)
The soundness of an exception for prehypnotic evidence has already been questioned in the scientific community.46 The proposed exception has also been vigorously attacked in the scholarly literature. (E.g., Mickenberg, Mesmerizing Justice: The Use of Hypnotically-Induced Testimony in Criminal Trials (1983) 34 Syracuse L.Rev. 927, 969-974.) In addition to the obvious hearsay problems presented by most types of prehypnotic evidence (see fn. 45, ante), the proposal raises constitutional issues of impairment of the right to confrontation and potential denial of a fair trial that may be different in degree but are not different in kind from those presented by the use of pretrial hypnosis in general. (See, e.g, Taylor, Scientific Interroga*429tion (1984) pp. 147-164; Mickenberg, op. cit. supra, 34 Syracuse L.Rev. at pp. 956-957; Beaver, Memory Restored or Confabulated by Hypnosis— Is It Competent? (1983) 6 U. Puget Sound L.Rev. 155, 195-202.)
We decline to render an advisory opinion answering those difficult questions in the case at bar. Respondent simply recites the holdings of the out-of-state cases without further analysis, and defendants have not briefed the matter at all. It will be time enough to grapple with such questions in a case in which they are presented as dispositive issues on a satisfactory record and after full briefing. We agree with the Michigan Supreme Court that the admissibility of prehypnotic evidence is a question that should be “reserved until raised on an adequate record in an appropriate case.” (People v. Gonzales (Mich. 1983) supra, 336 N.W.2d 751 (mod. of 329 N.W.2d 743).)
V.
We therefore apply the Shirley rule to the facts of this case. Under that rule we hold that the trial court erred in denying defendants’ two successive motions to exclude Judy’s testimony. (31 Cal.3d at pp. 66-67.) And because that testimony was virtually the sole incriminating evidence against each defendant, we find the error to be prejudicial under the Watson test. (Id. at p. 70.) In these circumstances we need not reach any other issues raised on appeal.
The judgments are reversed.
Bird, C. J., Broussard, J., Reynoso, J., and Grodin, J., concurred.
In the factual narrative that follows we have drawn from the transcripts of both trials. The testimony that each of the three principal witnesses gave in the Guerra trial was essentially the same as he or she gave in the Murkidjanian trial. In the few instances in which there was any significant difference, it will be noted.
She so testified at Guerra’s trial; at Murkidjanian’s trial she changed her story and denied that she returned the hug.
In his formal report of the interview, introduced as an exhibit to Guerra’s motion for new trial, Officer Granados reiterated Judy’s description of the events of the previous night; with regard to Murkidjanian’s actions her statement was similar to her trial testimony, with one significant omission—she made no mention whatever of penetration. On the form the officer identified the crime he was reporting as “Rape (Attempt).”
According to the police report, Judy told Officer Granados that Guerra “did not participate in the incident and said nothing during the event”; and that after Murkidjanian left, Guerra stayed with her and “asked her if she was okay and said he was sorry for what [Murkidjanian] had attempted to do.” At trial, Judy testified that at the time she was interviewed by Officer Granados her memory was fresh and she was trying to be truthful.
The following colloquy took place during cross-examination of Judy at the Murkidjanian trial:
“Q. [By defense counsel.] Did he tell you the reason to go to the hypnotist? A. Well, yes.
“Q. What did he tell you was the reason you should go? A. So that I could remember exactly what had happened.
“Q. In regards to whether or not there was penetration? A. I believe so.”
As noted in Shirley (31 Cal.3d at p. 57, fn. 35), this organization is a proprietary school in Los Angeles that teaches courses in hypnosis to police and other law enforcement personnel.
I.e., Murkidjanian, Guerra, and Logue.
Officer Davinroy testified as follows:
“Q. [By defense counsel.] Now, during the hypnotic session was there a time when the tape was not on? A. The tape recorder was always on, but it wasn’t always running.
“Q. Can you explain what you mean by that? A. Yes. As I seem to recall, Detective O’Brien kept going into hypnosis and I think there were periods of time when he didn’t realize that the tape had run out.
“Q. I see. As to the length of time the machine might have been off, we would need Detective O’Brien to give us the actual amount of time? A. I don’t think he would be able to do that.
“Q. Is that because he was hypnotized? A. Yes, sir, and people have a different perception of time when they are in hypnosis.”
In his report on the hypnotic session, introduced as an exhibit to Guerra’s motion for new trial, the officer estimated the level of the trance as “medium.” At trial the tape recording of the session was played in open court, and the transcript of the recording was furnished to the jury. Both the tape and the transcript were admitted into evidence and are part of the record on appeal.
Dr. Pennock testified only at the Murkidjanian trial.
Among the prehypnotic expectations that Officer Davinroy expressed were such statements as “We’ll get more information [by hypnosis] than we will if I was just talking to you normally”; “my only objective in being here conducting this session is to gain information that’s going to be used in a case”; “the information we gain is private; however, there is a possibility that the tapes would be reviewed in a court of law”; “from a prosecution standpoint we would love to get it so the jury could hear this”; and “In this particular case we’re interested in retrieving information regarding the type of assault.”
Dr. Rosenstein testified only on Guerra’s motion for new trial.
Other examples have included a decision in which we gave effect to a statutory rule that the courts had theretofore misconstrued (People v. Mutch (1971) 4 Cal.3d 389, 394 [93 Cal.Rptr. 721, 482 P.2d 633]) or had not definitively addressed (People v. Garcia (1984) supra, 36 Cal.3d at p. 549), and a decision that was our first occasion to apply certain binding precedents of the United States Supreme Court (People v. Sesslin (1968) 68 Cal.2d 418 [67 Cal.Rptr. 409, 439 P.2d 321], explained in People v. Groves (1969) 71 Cal.2d 1196, 1198 & fn. 1 [80 Cal.Rptr. 745, 458 P.2d 985]).
A recent example of a decision resolving a conflict in lower-court authority is People v. Beeman (1984) 35 Cal.3d 547, 556-560 [199 Cal.Rptr. 60, 674 P.2d 1318].
Even that exception is a limited one, in the sense that the normal rule of retroactivity always applies to “the individual defendant whose appeal is being adjudicated by the court” (People v. Bustamante (1981) supra, 30 Cal.3d 88, 102).
See, e.g., Berger v. California (1969) 393 U.S. 314 [21 L.Ed.2d 508, 89 S.Ct. 540]; McConnell v. Rhay (1968) 393 U.S. 2 [21 L.Ed.2d 2, 89 S.Ct. 32]; Arsenault v. Massachusetts (1968) 393 U.S. 5 [21 L.Ed.2d 5, 89 S.Ct. 35]; Roberts v. Russell (1968) 392 U.S. 293 [20 L.Ed.2d 1100, 88 S.Ct. 1291]; Witherspoon v. Illinois (1968) 391 U.S. 510, 523, footnote 22 [20 L.Ed.2d 776, 88 S.Ct. 1770].
For example, it pointed out that retroactive application of Winship affected only juvenile eourts, while the same applieation at Mullaney would invalidate the eonvictions of murderers.
In Johnson a five-member majority held that Fourth Amendment decisions in general, and Payton v. New York (1980) 445 U.S. 573 [63 L.Ed.2d 639, 100 S.Ct. 1371], in particular, are retroactive in the sense that they apply on direct appeal to all judgments not yet final. In Solem a different five-member majority held that a Sixth Amendment decision, Edwards v. Arizona (1981) 451 U.S. 477 [68 L.Ed.2d 378, 101 S.Ct. 1880], is prospective in the sense that it does not apply on habeas corpus to final judgments.
We expressed our preliminary view of Johnson in the second paragraph of footnote 10 of Donaldson, at page 38 of 35 Cal.3d.
The Courts of Appeal have also so held in a number of contexts. (See, e.g., People v. Hickman (1981) 127 Cal.App.3d 365, 369-370 [179 Cal.Rptr. 541]; People v. Faught (1981) 124 Cal.App.3d 848, 856-857 [177 Cal.Rptr. 637]; People v. Cooper (1979) 94 Cal.App.3d 672, 680 [156 Cal.Rptr. 646]; People v. Whittington (1977) 74 Cal.App.3d 806, 823-824 [141 Cal.Rptr. 742].)
We expressly disapproved the latter holding of Diggs in Shirley, pointing out that “the Kelly-Frye requirement is not fulfilled merely by evidence that one expert personally believes the challenged procedure is reliable; the court must be able to find that the procedure is generally accepted as reliable by the larger scientific community in which it originated. (Kelly, at pp. 30-32, 37 [of 17 Cal.3d].) It is obvious that no such finding could be made on the record in Diggs.” (31 Cal.3d at p. 54, fn. 32.)
Indeed, a casenote in an official publication of the California Department of Justice, the Peace Officer Law Report (Feb. 1981, p. 31), described Diggs as “the first published opinion in California on this controversial issue,” and pointed out that the California Supreme Court had several hypnosis cases pending before it but had not yet addressed the issue. The Peace Officer Law Report is distributed to criminal justice agencies throughout the state.
Obviously the police did not rely on Diggs or Colligan in the case at bar, as Judy was hypnotized before either was decided.
The cases are cited in Shirley. (31 Cal.3d at pp. 35-39.)
The cases rejecting hypnotically induced testimony are also cited in Shirley. (Id. at pp. 40-51.)
We therefore disapprove People v. Williams (1982) 132 Cal.App.3d 920 [183 Cal.Rptr. 498], which held to the contrary. As with the rule we adopted in Gainer (19 Cal.3d at p. 853), we do not reach the question whether the Shirley rule applies on collateral attack to cases now final.
The Collins opinion impliedly acknowledged that the rule it adopted was in effect partially retroactive when it observed in closing (ibid.) that “This Court is cognizant that its holding is not purely prospective . . . .” The implication was made express later in the same year by the Arizona Court of Appeals, which opined that the guidelines declared in Collins “might more properly be characterized as contemplating the application of a limited retroactive standard. Thus, the application of Collins would be prospective only in the sense that it would not be applied so as to invalidate convictions which have become final. On the other hand, it would apply to convictions which have not become final because of pending appeals. To that extent it would be given a limited retroactive effect in accordance with the general rule that when there is a change of law by judicial decision between the time of trial and appeal, the appellate court will dispose of the appeal according to the law prevailing at the time of the appellate disposition.” (State v. Young (App. 1982) 135 Ariz. 437 [661 P.2d 1138, 1140-1141].) The Arizona Supreme Court declined to review the decision.
Similarly, in Com. v. Nazarovitch (1981) 496 Pa. 97 [436 A.2d 170], the Supreme Court of Pennsylvania followed Mena and Mack in barring hypnotically induced testimony, and we relied on its reasoning in Shirley (31 Cal.3d at pp. 50-51). In Com. v. McCabe (1982) 303 Pa.Super. 245 [449 A.2d 670], the defendant was charged with a double murder and an assault with a deadly weapon, an eyewitness having identified him as the assailant only after being hypnotized. Both the crimes and the hypnosis took place prior to Nazarovitch, but the court nevertheless applied the rule of the latter case in affirming a pretrial order suppressing the tainted testimony.
As we observed in Shirley (31 Cal.3d at pp. 48-50), in Polk v. State (1981) 48 Md.App. 382 [427 A.2d 1041], the Maryland intermediate appellate court had held that Frye was controlling but that the question of the scientific consensus on hypnosis was still open. Collins answered that question.
Neither article is apparently yet in print. Respondent states that both are “soon to be published,” but tells us where only one is to appear: the Spiegel article will be published in 1984 in “Advances in Forensic Psychology and Allied Disciplines,” but no further citation is given. Respondent is silent as to when and where the Relinger article will appear. He appends copies of both to his supplemental brief.
See, e.g., H. Spiegel, Hypnosis and Evidence: Help or Hindrance? (1980) 347 Annals N.Y.Acad.Sci. 73 (cited in Shirley at p. 65, fn. 49, of 31 Cal.3d); Kroger & Doucé, Hypnosis in Criminal Investigation (1979) 27 Internal.J. Clinical & Experimental Hypnosis 358 (cited in Shirley, ibid.)-, Schafer & Rubio, Hypnosis to Aid the Recall of Witnesses (1978) 26 Internal.!. Clinical & Experimental Hypnosis 81 (cited in Shirley at p. 68, fn. 55, of 31 Cal. 3d).
By its frequent references to Bernard Diamond, the article somewhat misleadingly implies that Dr. Diamond is the sole or principal opponent of the use of posthypnotic testimony. As Shirley made clear, however, Dr. Martin T. Orne shares many of Dr. Diamond’s concerns on this topic, as do numerous other experts in the field. (See references cited in fns. 46, 48, 49, 51, and 52 of Shirley, at pp. 64-66 of 31 Cal.3d.) Indeed, we carefully explained in Shirley (id. at p. 63, fn. 45) that “By relying primarily on Drs. Diamond and Orne, of course, we do not mean to denigrate the contributions of many other experts in hypnosis whose writings we have also consulted (e.g., Ernest R. Hilgard, Ph.D., director of the Stanford University laboratory of hypnosis research), some of whom we cite hereinafter. On the present issue, however, the majority are in full agreement with the essential findings and conclusions of Drs. Diamond and Orne.”
In the typical study a group of subjects is exposed to certain material and their ability to recall it is tested; thereafter their recall is tested again while half of the group is in a hypnotic trance and the other half is not. If the hypnotized subjects recall more than the control subjects, the difference is deemed evidence of hypnotic hypermnesia.
See footnote 29, ante. If anything, the Relinger article in fact supports much of Shirley. The most popular method of memory enhancement used by law enforcement hypnotists is regression, a procedure described by Dr. Relinger as “an elaborate series of suggestions to hypnotized subjects attempting to convince them that they are actually reliving the event.” Dr. Relinger rejects this procedure as unreliable because of “The inevitable confabulation in addition to the excessive demand characteristics created by the elaborate regression suggestions . . . .” (“Demand characteristics,” in this context, are the verbal and nonverbal cues inherent in every hypnotic session.) He further rejects the “television technique” often used by law enforcement hypnotists (see Shirley, 31 Cal.3d at pp. 57 and 59, fn. 36) “in which subjects are told that their memory is like a videotape on which everything they have experienced is recorded”; such techniques, according to Dr. Relinger, “seem unduly suggestive and place unnecessary demands” on the person under hypnosis. In the case at bar Officer Davinroy used a variation of the “television technique” on Judy.
From these and other studies Dr. Wagstaff concluded in another article that “there is no scientific support for the view that evidence obtained under hypnosis is more truthful or necessarily more accurate than evidence obtained by procedures which do not employ hypnotic induction.” (Wagstaff, Hypnosis and the Law: A Critical Review of Some Recent Proposals, 1983 Crim.L.Rev. 152, 157.)
The reach of the study goes beyond its title: the authors note that although they used highly hypnotizable subjects, “Given the demand characteristics of the investigative hypnotic context, however, even witnesses and victims who are not especially responsive to hypnosis may be vulnerable also to such memory contaminants. Many individuals with low susceptibility to hypnosis have imagery that is as vivid as that of highly responsive individuals.” (Fns. omitted, ibid.)
The latter conclusion is further supported by a more recent study which concluded, inter alia, that hypnosis strongly increases the subject’s confidence in the accuracy of his recall of both true and false memories. (Sheehan & Tilden, Effects of Suggestibility and Hypnosis on Accurate and Distorted Retrieval from Memory (1983) 9 J. Experimental Psychology: Learning, Memory, & Cognition 283.)
Fortunately, no miscarriage of justice occurred: in one of the cases reported the jury was unable to agree, and in the other the charges were dismissed when the victim departed from the state.
For example, the subject’s own preconceptions about hypnosis may play a major role: if, as often occurs, he believes hypnosis will allow him to retrieve memories he could not otherwise recall, this belief itself becomes an effective prehypnotic suggestion; he will strive to comply with it while in the trance, even at the cost of altering his true recall. (Id. at pp. 175-176.) In addition, even standard hypnotic technique can contaminate the subject’s memory. The hypnotist commonly encourages the subject to proceed with his narration by making such seemingly innocuous remarks as “Good,” “Fine,” or “You’re doing well.” To the subject, however, these remarks may take on much greater significance, suggesting to him that the hypnotist accepts his tentative memories as objective truths, and this misperception may in turn persuade the subject they are true. (Id. at p. 176.) Moreover, the subject may misinterpret even a slight and unintended decrease in the frequency of such remarks as meaning that the hypnotist wants “something else or something more,” and in an effort to please he may produce unreliable or false information. (Ibid.) Lastly, the “television” or “videotape” technique of police hypnotists inevitably pressures the subject to produce additional details and to believe they are true, yet his subsequent testimony to that effect “will be based on memories created during hypnosis, which may be at gross variance with both prehypnosis recollection and the actual facts.” (Id. at p. 200.)
For example, hypnosis may produce a dramatic increase in the amount of detail the witness reports, or in the emotion which accompanies that report. Even though they may be artifacts of the hypnotic process wholly unrelated to the facts, these additional details or emotion tend to make the witness more credible to lay observers such as a jury. (Id. at pp. 193-194.)
Even when hypnosis is restricted to this purpose, Dr. Orne is of the opinion that its “use can be justified, however, only in cases where a defendant has not been identified to the subject, where there has not been widespread publicity involving speculations about the perpetrator, and where law enforcement officials do not have compelling beliefs about details of what actually transpired.” (Id. at p. 205.) And even in these limited cases, Dr. Orne would still require strict compliance with a newly expanded set of guidelines that takes almost five full pages of his article to explain. (Id. at pp. 205-210.)
Shortly before Shirley, New Mexico so held as well. (State v. Beachum (App. 1981) 97 N.M. 682 [643 P.2d 246, 249-255].)
Shortly before Shirley, Wyoming likewise so held. (Chapman v. State (Wyo. 1982) 638 P.2d 1280, 1282-1285.)
E.g., U.S. v. Valdez (5th Cir. 1984) 722 F.2d 1196, 1202; State ex rel. Collins v. Superior Court, etc. (Ariz. 1982) supra, 644 P.2d 1266, 1272-1273, 1288-1295; People v. Quintanar (Colo.App. 1982) 659 P.2d 710, 712-713; State v. Collins (Md. 1983) supra, 464 A.2d 1028, 1043-1044; People v. Gonzales (Mich. 1982) supra, 329 N.W.2d 743, 747; People v. Hughes (N.Y. 1983) supra, 453 N.E.2d 484, 494-495; State v. Peoples (N.C. 1984) supra, 319 S.E.2d 177, 185-186, 188; Com. v. Nazarovitch (Pa. 1981) supra, 436 A.2d 170.
As discussed above (Part III C, ante), its old adherents also remain steadfast: the courts that adopted the rule before Shirley have since applied it to cases pending on appeal at the time of its adoption.
The federal court limited its holding to the particular type of posthypnotic testimony before it, i.e., testimony in which a hypnotized witness identifies for the first time a person he knew was already under suspicion. The court also based its holding on the Federal Rules of Evidence rather than the Frye test. (Id. at pp. 1200-1201.) Valdez was thus a cautious step, but it was a first step nonetheless; all prior federal appellate decisions on the topic allowed posthypnotic testimony into evidence subject to certain “safeguards.” (See Shirley, 31 Cal.3d at pp. 36-37 & fn. 19.)
Such evidence might be presented in a wide variety of forms: e.g., a police report of a field interview of the witness immediately after the crime; a subsequent interrogation of the witness by a detective at the police station for the purpose of advancing the investigation of the crime; a formal deposition of the witness by a prosecutor at the latter’s office for the purpose of preserving his testimony prior to hypnosis; trial testimony by the witness as to facts he claims to have remembered before hypnosis; and trial testimony by other witnesses, including law enforcement authorities, as to statements assertedly made by the percipient witness before hypnosis.
Thus Dr. Orne and his colleagues warn that it is unclear “whether a witness can distinguish between recollections made prior to hypnosis and those reported during and after hypnosis . . . .” (Orne et al., Hypnotically Induced Testimony, supra, at p. 203.) The difficulty is that “the sources of memories become confounded, and no instructions or suggestions can reliably prevent this from occurring in real-life situations.” {Id. at p. 200.) The confusion is compounded, moreover, by the tendency of hypnotized persons to deny the effects of their hypnotic experience: the same researchers report the phenomenon that “witnesses who have been hypnotized often assert later that their new recollections preceded rather than followed the hypnotic session,” and that even subjects who “had absolutely no recollection prior to hypnosis” may “insist that they actually remembered the event or detail before being hypnotized” but simply “did not talk about it.” {Ibid.) Finally, the authors observe that the artificial increase caused by hypnosis in the witness’s confidence in his recall, whether true or false, affects prehypnotic no less than posthypnotic testimony. {Id. at p. 203.) In short, it appears from their entire article that Dr. Orne and his colleagues are of the opinion that the testimony of a witness as to his prehypnotic recollection of events is not reliable and that its admission substantially impairs the ability to cross-examine that witness.