People v. Wright 45 Cal. 3d 1126

Opinion

LUCAS, C. J.

In this case we consider whether the trial court erred in rejecting five special jury instructions which defendant requested relating to the accuracy of eyewitness identifications. Defendant was convicted of armed robbery and related offenses in connection with a robbery at gunpoint by several masked men. The sole evidence against him at trial was eyewitness identification.

We conclude that the court correctly declined to give four of the five requested instructions. We also conclude that the court erred in failing to *1132give an instruction listing the factors the jury could consider in evaluating eyewitness identifications, but we find the error was harmless.

I. Facts

Defendant was charged with participating in the armed robbery of 11 employees of a wholesale beverage company in the company’s warehouse and office building. His defenses were mistaken identification and alibi. The jury found him guilty of nine counts of armed robbery, two counts of attempted armed robbery, and one count each of assault with a deadly weapon and false imprisonment.

During the robbery, a group of men armed with handguns and wearing stocking masks burst into the company’s warehouse, forced the employees to lie on the floor, and demanded their money. Some of the robbers then proceeded to the main office where they ordered the office employees to the floor and again demanded money. In addition, they forced one employee to open the company safe, and removed money from it. They also demanded jewelry from the victims. The robbers left the scene about 20 minutes after they arrived.

Three of the eleven employees present at the robbery identified defendant as one of the robbers. Stephanie Sung, the employee who opened the safe, was able to identify two of the robbers. As she went to the safe and opened it, she came “face to face” with one of them. She was able to observe this man’s face through his black stocking mask, which, she testified, “wasn’t very tight” and “didn’t really smash his features.” She also saw the face of a second robber as she returned from opening the safe. Later on the day of the robbery, she selected from a photo spread two photographs as resembling the robbers whose faces she had seen. One of these was defendant’s photo, which Sung testified looked like the robber she saw as she opened the safe. She testified that at a lineup a week later she recognized defendant and was “75 percent sure” he was the robber she had seen at the safe, but she did not sign her lineup card, believing she had to be “a hundred percent sure” in order to sign it.1 During Sung’s testimony, at the prosecutor’s request, defendant put a stocking mask over his face. Based on this in-court demonstration, Sung positively identified defendant as the robber she had seen at the safe.

Witness Erica Albertsen saw a man wearing a black stocking mask run toward her through the office door. The man grabbed her shoulders, lifted *1133her up, and pushed her to the floor. She stated his mask did not hide his face, and she could see “very clearly” through the mask. She identified defendant as the robber she saw. She testified she had not seen a photo of defendant in the postrobbery photo spread, and that at the police lineup conducted later she was “amazed” to see defendant. At the lineup, she immediately recognized defendant as the robber she had seen. Further, she explained that her description of defendant at the preliminary hearing was from memory, given while defendant was hidden from her view.

Peter Marino also testified that defendant was one of the robbers, although he made a weaker identification. He picked defendant’s photograph from the photo spread, saying, “this man [defendant] was in the warehouse that morning.” At trial, he said that he could not specify which one of the group of robbers was defendant, but testified, “I believe [defendant] was in the robbery that day.”

The eight other victims of the robbery testified that they were unable to say whether or not defendant was one of the robbers. The testimony of all the eyewitnesses will be discussed more fully where relevant below.

Defense counsel used a model to demonstrate the effect of the stocking masks. The model put on several masks, giving the jury an opportunity to observe how they distorted or otherwise affected the wearer’s facial features. As noted, when the prosecutor had the defendant model a stocking mask, Stephanie Sung was able to identify him positively.

Defendant did not testify at trial. His wife testified that on the day of the robbery he was with her all day, sleeping most of the morning, and then decorating a bar for a birthday party. Defendant’s sister-in-law testified that she had also been with defendant and his wife decorating for the party. Their testimony was uncorroborated and somewhat weak.2

Defendant was tried together with a codefendant charged with participating in the same robbery. The case against the codefendant was also based on eyewitness identification testimony. The presentation of evidence took about eight days. The jury, after deliberating less than one day, found defendant guilty on all counts, but it was unable to reach a verdict as to the *1134guilt of the codefendant. Defendant appealed on the sole ground that the trial court committed reversible error by refusing to give his five requested special instructions on eyewitness identification.

II. Discussion

A. Defendant’s Special Instructions Nos. 1 and 4

Requested instructions Nos. 1 and 4 state that the prosecutor has the burden of proof on the issue of identity.3 They essentially duplicate each other; more importantly, they also duplicate former CALJIC No. 2.914 (prosecutor’s burden of proving beyond a reasonable doubt the accuracy of eyewitness identifications) which the court gave. (Post, at p. 1149.) Defendant argues that his instruction No. 1 would additionally inform the jury that the identity of the criminal is an element of the crime; but this point is plainly implied by the statement in former CALJIC No. 2.91 that the prosecution has the burden of proving such identity beyond a reasonable doubt.

Defendant also notes that instruction No. 4 would tell the jury he need not prove his innocence or another’s guilt; but the court stated this rule in CALJIC No. 2.90, which defines the presumption of innocence and the prosecutor’s general burden of proof beyond a reasonable doubt. Defendant’s special instructions Nos. 1 and 4 are thus repetitious of instructions already given, and the trial court correctly refused them on this ground. (People v. Martinez (1987) 191 Cal.App.3d 1372, 1378-1379 [237 Cal.Rptr. 219]; People v. McCowan (1978) 85 Cal.App.3d 675, 679-680 [149 Cal.Rptr. 611].)

*1135B. Defendant's Special Instruction No. 2

Defendant’s second proposed instruction lists certain specific items of evidence introduced at trial, and would advise the jury that it may “consider” such evidence in determining whether defendant is guilty beyond a reasonable doubt.5 The court refused to give this instruction because it is argumentative, i.e., it would invite the jury to draw inferences favorable to the defendant from specified items of evidence on a disputed question of fact, and therefore properly belongs not in instructions, but in the arguments of counsel to the jury.

The court ruled correctly. In People v. McNamara (1892) 94 Cal. 509 [29 P. 953], a defendant charged with robbery was identified at trial by the victim. The court refused to give a proposed defense instruction telling the jury that in determining whether it had a reasonable doubt of the identity of the assailant it should “consider” specific evidence listed in the instruction, for example, that the victim was the sole witness to identify the defendant, that the victim was a stranger to the city, and the victim’s “condition of sobriety or insobriety”; and that the defendant did not flee the city after the crime, but expressed his willingness to help recover the property and capture the robbers. (Id., at pp. 513-514.) We held the instruction was properly rejected as argumentative. We disapproved of “the common practice [of] selecting] certain material facts, or those which are deemed to be material, and endeavoring to force the court to indicate an opinion favorable to the defendant as to the effect of such facts, by incorporating them into instructions containing a correct principle of law,” and we explained, “An instruction should contain a principle of law applicable to the case, expressed in plain language, indicating no opinion of the court as to any fact in issue.” (Id., at p. 513; accord, People v. Hill (1946) 76 Cal.App.2d 330, 342 [173 P.2d 26]; see also People v. Slocum (1975) 52 Cal.App.3d 867, 893 [125 Cal.Rptr. 442].)6

*1136The foregoing rule was not undermined by the 1934 constitutional amendment permitting a trial court to “make such comment on the evidence and the testimony and credibility of any witness as in its opinion is necessary for the proper determination of the cause.” (Cal. Const., art. VI, former § 19, now § 10.) We deal here with an instruction, not a judicial comment; although such a comment may address matters of fact, an instruction may not.7 “[I]t is not a matter of law for the judge to say that certain evidence might give rise to a reasonable doubt as to the affirmative of an issue required to be proven by the prosecution. That is a comment on the evidence and any such comment should be identified as such.” (People v. Castellano (1978) 79 Cal.App.3d 844, 858 [145 Cal.Rptr. 264].) Moreover, even if defendant had characterized his request as a proposed judicial comment on the evidence rather than an instruction, it would still have been improper. “Judicial comment should be temperate rather than argumentative and the trial court must avoid engaging in partisan advocacy.” (People v. Cook (1983) 33 Cal.3d 400, 408 [189 Cal.Rptr. 159, 658 P.2d 86].) Judicial comment “should, in the interests of justice, not be limited within an area selected by the defendant, nor given in language of his choosing.” (Castellano, supra, 79 Cal.App.3d at p. 858.)

Defendant’s instruction No. 2 follows the form of a requested instruction quoted in People v. Sears (1970) 2 Cal.3d 180, 189-190 [84 Cal.Rptr. 711, 465 P.2d 847]. The Sears instruction would have told the jury it could consider evidence such as “[t]he lack of motive” and “defendant’s prior consumption of alcoholic beverage” in determining whether there was a reasonable doubt as to the existence of premeditation and deliberation. (Id., at p. 189.) We held in Sears that the defendant was entitled to an instruction relating certain evidence brought out at trial to the issue of premeditation and deliberation. But the quoted instruction was not intended to serve as a model; indeed, we warned in Sears that the instruction is “defective in form in some respects” and emphasized that on retrial the court should give “appropriate instructions, if they are requested.” (Id., at p. 190, italics added; see People v. Whittaker, supra, 41 Cal.App.3d at p. 308.)

In addition, defendant mistakenly argues that Sears supports his position by its statement that a defendant “has a right to an instruction that directs *1137attention to evidence from a consideration of which a reasonable doubt of his guilt could be engendered. (People v. Granados [(1957)] 49 Cal.2d 490, 496 [319 P.2d 346].)” (2 Cal.3d at p. 190.) This statement is taken out of context. Neither it, nor the broader observation in Sears that “[a] defendant is entitled to an instruction relating particular facts to any legal issue” (ibid.), abrogated sub silentio the well settled rule against argumentative instructions on a disputed question of fact. “ ‘None of the cases cited in Sears says, and Sears itself does not say, that upon request the judge must direct the attention of the jury to specific testimony and tell the jury it may look to that testimony for the purpose of forming a reasonable doubt on an issue.’ ” (Castellano, supra, 79 Cal.App.3d at p. 858.)

As appears from an analysis of the full Sears opinion (supra, 2 Cal.3d 180), and from the sources it relied on, the language quoted by defendant refers only to a defendant’s right to an instruction that “pinpoint[s] the theory of the defense.” (Granados, supra, 49 Cal.2d at p. 496.) In a proper instruction, “[w]hat is pinpointed is not specific evidence as such, but the theory of the defendant’s case.” (People v. Adrian (1982) 135 Cal.App.3d 335, 338 [185 Cal.Rptr. 506], original italics.)

The difference between an instruction that pinpoints the crux of the defense and one that improperly implies certain conclusions from specified evidence is illustrated by People v. Wilson (1929) 100 Cal.App. 428 [280 P. 169], a case relied on in Sears.8 The defendant in Wilson, charged with robbery, put on an alibi defense. The court instructed generally that the prosecution must prove every material fact beyond a reasonable doubt, and defined reasonable doubt by the statutory instruction set out in Penal Code section 1096 (CALJIC No. 2.90). The defendant specially requested additional instructions that explained the alibi defense, relating it to the prosecution’s burden of proof, and advising the jury that if it believed the defendant was not present when the crime was committed, he should be acquitted. The court refused these instructions.

On appeal from his conviction, the defendant in Wilson relied on earlier cases holding such alibi instructions proper. The People argued that because of the then-recent enactment of Penal Code section 1096a, rejection of the alibi instruction was not error. Section 1096a declares that if the court gives the section 1096 statutory instruction on reasonable doubt, “no further instruction . . . defining reasonable doubt need be given.” The Court of Appeal reversed the conviction, distinguishing between instructions merely *1138“defining” reasonable doubt and those “covering a defense as shown by the evidence in the case from which reasonable doubt of the guilt of the defendant may be inferred.” (Id., at p. 432, italics added.) The defendant’s requested instruction was in the latter category, for it pinpointed the theory of the defense, namely alibi, and charged the jury on how to relate the evidence of that defense to the prosecution’s general burden of proving guilt beyond a reasonable doubt. The alibi instruction upheld in Wilson was subsequently formalized as CALJIC No. 4.50, and was given in the present case. (Post, at p. 1150.)

A similar evolution occurred with respect to a special instruction on the mistaken identification defense. In People v. Gomez (1972) 24 Cal.App.3d 486 [100 Cal.Rptr. 896], the defense was both alibi and mistaken identification. The trial court gave the alibi instruction of CALJIC No. 4.50, but refused to give a requested instruction explaining that the prosecution has the burden of proving not only that the offense was committed but that it was the defendant who committed it; that the jury must be satisfied beyond a reasonable doubt of the accuracy of the identification of the defendant before it may convict; and that if the circumstances of the identification are not convincing beyond a reasonable doubt, the jury must find the defendant not guilty. The court rejected this identification instruction on the ground the issue was covered by the general statutory instruction on proof beyond a reasonable doubt as set out in Penal Code section 1096 and CALJIC No. 2.90. The Court of Appeal held this refusal error, relying on the Wilson-Sears language quoted ante at page 1137. (Id., at p. 490.)

The Gomez instruction on the burden of proof of identification was formalized as CALJIC No. 2.91. In the present case the court gave the Gomez instruction, thus pinpointing the defense theory of mistaken identification and charging the jury on how to relate the evidence of that defense to the prosecution’s general burden of proving guilt beyond a reasonable doubt. (See post, at p. 1149.) Defendant’s further requested instruction No. 2 was properly rejected as argumentative. Numerous decisions have held similar instructions properly refused. (E.g., People v. Kaiser (1980) 113 Cal.App.3d 754, 764-766 [170 Cal.Rptr. 62]; People v. Guy (1980) 107 Cal.App.3d 593, 602-604 [165 Cal.Rptr. 463]; People v. Watson (1979) 89 Cal.App.3d 376, 385-387 [152 Cal.Rptr. 471]; People v. Smith, supra, 67 Cal.App.3d at pp. 48-49; People v. Whittaker, supra, 41 Cal.App.3d at pp. 307-309.)

C. The Eyewitness Identification “Factors” Instruction

1. Failure to Give the Requested Instruction Was Error

Defendant’s requested instruction No. 3 lists certain factors, supported by the evidence in the case, that are relevant to the jury’s evaluation *1139of eyewitness identifications. It focuses on three main areas: 1) the witness’s ability to observe the person being identified at the time of the crime; 2) the reliability of the witness’s subsequent identification of the defendant; and 3) inconsistent or erroneous identifications the witness may have made before trial.9 We hold it was error, in this case, to reject the requested instruction.

Defendant’s special instruction derives from a well-known model instruction originally promulgated in United States v. Telfaire (1972) 152 App.D.C. 146 [469 F.2d 552, 558-559].10 The first California decision to discuss the Telfaire instruction was People v. Guzman (1975) 47 Cal.App.3d 380, 386-387 [121 Cal.Rptr. 69].11 Guzman held that although the Telfaire instruction was probably “too long and argumentative” as written, it “at least presents the basic problem and could have been modified to eliminate the faults contained.” (Id., at p. 386.)

After Guzman, various courts held that a refusal to give a requested Telfair e-Guzman instruction either was not error, or was harmless because the trial court gave general instructions on eyewitness identifications (usually CALJIC Nos. 2.20 and 2.91). (E.g., People v. Boothe (1977) 65 Cal.App.3d 685, 689-690 [135 Cal.Rptr. 570] [rejection of instruction *1140harmless under Guzman]; People v. Hurley (1979) 95 Cal.App.3d 895, 900-901 [157 Cal.Rptr. 364] [no error in rejecting a lengthy requested instruction; Telfaire-Guzman instruction or other detailed instruction not required].) We first discussed the Telfaire-Guzman instruction in People v. Hall (1980) 28 Cal.3d 143 [167 Cal.Rptr. 844, 616 P.2d 826], There, the trial court refused to give a proposed instruction identical to the instruction requested in Guzman. (Id., at pp. 158-159.) Some of the factors highlighted in the requested instruction were inapplicable to the evidence in the case, so it was not error to refuse to give the instruction as written. We held, however, that the court’s refusal to delete the inappropriate factors from the instruction and give a properly tailored instruction was error. (Id., at p. 159.) We found the error harmless because the court gave CALJIC Nos. 2.20 and 2.91, but declared, “In the future, the trial courts should consider and give appropriate instructions involving reasonable doubt and eyewitness identification.” (Id., at pp. 159-160.) We stated, “it is error to refuse to give an instruction requested by a defendant which ‘directs attention to evidence from . . . which a reasonable doubt of guilt could be engendered.’ (People v. Sears [supra] 2 Cal.3d 180, 190 [84 Cal.Rptr. 711, 465 P.2d 847].) This applies with equal force to a refusal to give a requested instruction which deals with identification in the context of reasonable doubt.” (Id., at p. 159.)

Following Hall, supra, 28 Cal.3d 143, some courts continued to hold that general instructions on eyewitness identification rendered Guzman-type instructions superfluous. (E.g., People v. Sequeira (1981) 126 Cal.App.3d 1, 17-18 [179 Cal.Rptr. 249].) Beginning with People v. West (1983) 139 Cal.App.3d 606 [189 Cal.Rptr. 36], however, the Courts of Appeal have generally held it error, in appropriate cases, to refuse an instruction listing the factors the jury may consider in evaluating eyewitness identification testimony. (E.g., People v. Palmer (1984) 154 Cal.App.3d 79, 89 [203 Cal.Rptr. 474]; People v. Coates (1984) 152 Cal.App.3d 665, 670-671 [199 Cal.Rptr. 675]; People v. Brown (1984) 152 Cal.App.3d 674, 678 [199 Cal.Rptr. 680].) In People v. Yeats (1984) 150 Cal.App.3d 983, the court reviewed Guzman, Hall, and West, and concluded that “an unbroken string of authorities requires the giving of factually appropriate pinpoint jury instructions correlating the issues of identity and reasonable doubt.” (Id., at p. 990.)12

*1141In West, supra, 139 Cal.App.3d 606, the requested instruction listed eight factors for the jury to consider in determining whether there was reasonable doubt ás to the identification of the defendant as the perpetrator. (Id., at p. 609.) It would have instructed the jury to consider evidence relating to such factors as the witness’s opportunity to observe the alleged criminal act, the stress under which the witness made the observations, the cross-racial nature of the identification, and whether the witness had an uncorrected vision deficiency. The West instruction has been formalized, with the addition of several other factors (including the length of time between the alleged criminal act and the witness’s identification, and whether the witness had prior contacts with the alleged perpetrator) in CALJIC No. 2.92 (4th ed. 1987 pocket pt.), entitled “Factors to Consider in Proving Identity by Eyewitness Testimony.” This model instruction, with appropriate modifications to take into account the evidence presented at trial, will usually provide sufficient guidance on eyewitness identification factors. (CALJIC No. 2.92 is set out in the appendix.)

As we stated in McDonald, supra, “the defendant may be entitled to a special instruction specifically directing the jury’s attention to other evidence in the record—e.g., facts developed on cross-examination of the eyewitnesses—that supports his defense of mistaken identification and could give rise to a reasonable doubt of his guilt.” (37 Cal.3d at p. 377, fn. 24, italics added.) We now reach the issue we avoided in McDonald, the correct wording of such an instruction. We hold that a proper instruction on eyewitness identification factors should focus the jury’s attention on facts relevant to its determination of the existence of reasonable doubt regarding identification, by listing, in a neutral manner, the relevant factors supported by the evidence.

The instruction should not take a position as to the impact of each of the psychological factors listed. We disagree with the dissent’s suggestion that CALJIC No. 2.92 is “deficient” for failing to explain the effects of the enumerated factors. (Post, p. 1157.) An instruction that “explained” the influence of the various psychological factors would of necessity adopt the views of certain experts and incorporate the results of certain psychological studies while discounting others. It would require the trial judge to endorse, and require the jury to follow, a particular psychological theory relating to the reliability of eyewitness identifications. Such an instruction would improperly invade the domain of the jury, and confuse the roles of expert witnesses and the judge.

McDonald, supra, 37 Cal.3d 351, contrary to the dissent’s implication (post, p. 1158), does not justify such an “explanatory” instruction. In Mc*1142Donald, we endorsed only the use of expert testimony to put before the jury the scientific and psychological evidence as to factors believed relevant to an evaluation of eyewitness identifications. As explained in McDonald, the use of expert testimony may assist the jury in its task of evaluating the testimony and other evidence before it, without usurping the jury’s functions of determining the credibility and qualifications of the witnesses, and affording their testimony the appropriate weight. (37 Cal.App.3d at pp. 369-371.) Expert witnesses may be called by both parties, thus enabling the jury to hear both sides of the scientific theories regarding eyewitness identifications.13 Questioning can focus on the psychological factors most relevant to the particular circumstances of the case, thereby avoiding presenting the jury with potentially inaccurate overgeneralizations. Further, cross-examination of experts can reveal the weaknesses in their testimony and theories, as well as the underlying assumptions on which their conclusions are based.

Unlike a jury instruction, expert testimony is not binding on the jury. In McDonald, we observed, “the jury remains free to reject [the expert testimony] entirely after considering the expert’s opinion, reasons, qualifications, and credibility. Indeed, the Penal Code commands (§ 1127b) that an instruction so informing the jury be given in any criminal trial in which expert opinion evidence is received.” (37 Cal.3d at p. 371, italics added, fn. omitted.) CALJIC No. 2.80 tells the jury in part, “You are not bound to accept an expert opinion as conclusive, but should give to it the weight to which you *1143find it to be entitled. You may disregard any such opinion if you find it to be unreasonable.”

In contrast, an instruction incorporating a particular expert’s opinion would deprive the jury of its independence in judging the weight to be given to such expert opinion. The dissent can cite no authority, and we have found none, to support endorsing an expert opinion, via jury instruction, on a matter of ongoing scientific debate. (See fn. 13, ante, and cases cited.)

We conclude that the listing of factors to be considered by the jury will sufficiently bring to the jury’s attention the appropriate factors, and that an explanation of the effects of those factors is best left to argument by counsel, cross-examination of the eyewitnesses, and expert testimony where appropriate.14 The instruction should list the applicable factors in a neutral and nonargumentative instruction, thus effectively informing the jury without improperly invading the domain of either jury or expert witness. It should list only factors applicable to the evidence at trial, and should refrain from being unduly long or argumentative.

In People v. Martinez, supra, 191 Cal.App.3d 1372, the court gave CALJIC No. 2.92, after first providing defense counsel an opportunity to suggest additional factors to supplement those listed in the standard instruction. (Id., at p. 1383.) This procedure will generally be a proper one. In Martinez, the defendant declined to add any factors to CALJIC No. 2.92. (Ibid.) The Court of Appeal held that it was not error for the court to reject an additional and lengthy “Guzman-like instruction” requested by the defendant, because “the factors and criteria set forth in [the requested instruction] have been sufficiently incorporated in CALJIC No. 2.92.’” (Ibid.) The court correctly noted, “The concern of the court in Guzman was that the trial court instruct on the ‘factors that affect eyewitness identification testimony, as they relate to reasonable doubt.’ [Citation.] CALJIC No. 2.92 clearly focused the jury’s attention on such factors.” (Id., at p. 1384.)

*1144CALJIC No. 2.92 or a comparable instruction should be given when requested in a case in which identification is a crucial issue and there is no substantial corroborative evidence. (See McDonald, supra, 37 Cal.3d at p. 377 [eyewitness expert testimony appropriate “[w]hen an eyewitness identification of the defendant is a key element of the prosecution’s case but is not substantially corroborated by evidence giving it independent reliability,” and in such a case the defendant also may be entitled to a special instruction on eyewitness identification factors].)

In this case, defendant’s requested instruction No. 3, like CALJIC No. 2.92, avoids being “too long and argumentative.” Although based on the Telfair e-Guzman instruction, it is a much condensed version of that charge; and rather than drawing factual inferences favorable to defendant from specific items of testimony, it “merely intone[s] considerations any jury must make in evaluating any eyewitness identification.” (People v. Guy, supra, 107 Cal.App.3d 593, 603.) As drafted in the present case, the instruction does not err by referring to factors unsupported by the evidence.15 (Cf. People v. Hall, supra, 28 Cal.3d 143, 159.) Therefore, we hold the court erred in refusing to give defendant’s special instruction No. 3.

2. The Error Was Harmless

Although the court erred in not giving defendant’s requested “factors” instruction, the error was not prejudicial. This error requires reversal only if “the court, ‘after an examination of the entire cause, including the evidence,’ is of the ‘opinion’ that it is reasonably probable that a result more favorable to [defendant] would have been reached in the absence of the error.” (People v. Watson (1956) 46 Cal.2d 818, 836 [299 P.2d 243].)

Our consideration of the entire record in this case leads us to conclude that it is not reasonably probable that had the requested instruction been given, the jury would have reached a verdict more favorable to defendant. We base this holding on (a) the overall strength of the evidence; (b) the fact that factors relating to the reliability of the eyewitness identifications were brought to the jury’s attention by (i) cross-examination, (ii) opening and closing arguments of counsel, and (iii) the jury instructions *1145given; and (c) the absence of any indication that the jury was uncertain or confused.16

(a) Overall Strength of the Evidence

The evidence as a whole was sufficiently convincing to warrant sustaining the guilty verdict.17 As noted, three of the eleven victims identified defendant as one of the robbers. Although there was no evidence corroborating the eyewitness identifications, neither did any eyewitness testify at trial that defendant was not one of the robbers, as specifically occurred in McDonald, supra, 37 Cal.3d 351. (In McDonald, an eyewitness, the same race as defendant, testified she had no doubt that defendant was not the perpetrator. [Id., at pp. 359-361.]) The witnesses in this case who were unable to recognize defendant were for the most part unable to describe the faces of any of the robbers.18 On this basis the facts of the present case are distinguishable from cases such as McDonald, supra, and West, supra, 139 Cal.App.3d 606, 608, in which at least one eyewitness testified unequivocally that the defendant was not the perpetrator. Indeed, we stated explicitly in McDonald that our finding of prejudicial error was based in part on the fact that the evidence as to identity “was close, given the potential weaknesses in the prosecution’s testimony and the presence of both eyewitness and alibi testimony favorable to the defense.” (37 Cal.3d at p. 376, italics added, fn. omitted.)

In addition, in this case, defendant’s alibi evidence was not strong and could reasonably have been discounted by the jury. (See ante, fn. 2.) This also distinguishes the present case from McDonald, in which the alibi evidence was substantially stronger. In McDonald, six witnesses testified that the defendant had been in Alabama visiting his family for several weeks, including on the day that the California murder of which he was accused took place. Postcards mailed by defendant while en route to Alabama, and telephone bills of defendant’s Alabama relatives for the relevant dates, were introduced as corroborative evidence as to his location. (37 Cal.3d at p. 360.)

*1146(b) Factors Affecting Reliability of Eyewitness Identification Testimony

The factors listed in defendant’s requested instruction (relating to the circumstances of the original observation, circumstances of subsequent identifications, and other inconsistent identifications) were put before the jury at trial by means of several vehicles: cross-examination, counsel’s arguments,19 and the instructions the court gave.

(i) Cross-examination

The 11 eyewitnesses to the robbery testified over the course of 5 days. Counsel for defendant and codefendant interrogated the witnesses extensively on their ability to see and identify the robbers during the robbery, and on the circumstances of their subsequent identifications. They succeeded in thoroughly highlighting the factors relevant to the reliability of the eyewitness identifications.

Cross-examination of the three witnesses who identified defendant focused on the factors listed in defendant’s requested instruction. For example, Peter Marino, who identified defendant at a photo lineup on the day of the robbery, admitted on cross-examination that at the robbery he could not see the robbers’ faces very well,20 and that he could not distinguish their features through the masks. He candidly admitted, “you don’t get a full look at everybody and you’re scared,” and acknowledged the distraction *1147caused by other employees screaming and yelling. Marino also conceded that, at defendant’s preliminary hearing, he had testified that he saw defendant in court the week before the hearing, and that defendant at that time did not appear to be the same man whose photograph he had selected. In addition, defense counsel impeached Marino by reading from his testimony at the codefendant’s preliminary hearing, in which Marino, when asked to select from a group of photographs the photo he had selected at the lineup on the day of the robbery, instead picked a photograph of someone other than defendant. Finally, Marino stated that his in-court identification of defendant was based on the photograph he had selected at the original photo spread.

Stephanie Sung, who opened the safe for the robbers, stated she came face to face with defendant, and that she could see his face despite her extreme nervousness. At trial, after defendant pulled a stocking mask over his face, she positively identified him as one of the robbers. Sung stated that based on that in-court view of defendant, she believed he was the robber she had seen while opening the safe. On cross-examination, she admitted that when she had selected defendant’s photo from a photo layout after the robbery, she had said only that the man in the picture “looked like” the robber, and had not stated defendant actually was the robber. Further, at the subsequent live lineup, she had not been sure enough of her identification to pick out defendant. She stated she had been 75 percent certain at the lineup, but at trial was 100 percent certain defendant was one of the robbers she saw.

The third witness, Erica Albertsen, also made an in-court identification of defendant. She testified that she had identified him in a police lineup seven to ten days after the robbery. Defense counsel questioned her closely on the extent of her opportunity to observe the robber. Albertsen testified she could see the robber’s facé as he came into her office, ran around the back of her chair, and then pushed her to the floor. She admitted being “scared and startled.” Defense counsel, reading from the preliminary hearing transcript in which Albertsen had stated the robber’s height was “five-seven,” noted that at trial she changed her description to “five-eight, five-nine.” Also, at the preliminary hearing Albertsen had described the robber’s skin color as “medium,” whereas at trial she changed it to “lighter than medium.” In addition, she testified that she had failed to give a complete description of the robber to the police inspector or anyone else before defendant’s preliminary hearing. She stated that this was because no one had asked her for such a description up to that point.

Finally, she admitted she had not selected defendant’s photo from the photo lineup, but insisted his photo had not been in that lineup. From the evidence, it is uncertain whether Albertsen was shown defendant’s photo. *1148The inspector who originally displayed the photo spreads gave conflicting testimony about whether he showed defendant’s photo to all the victims.

In addition, the eight employees who could not identify defendant all testified. Most of them stated that they were unable to discern or describe the features of the robbers. Their cross-examination focused on the factors contributing to their inability to identify defendant, such as poor eyesight, the stress of the event, their fear, concealment of the robbers’ faces by masks, and the fact that the witnesses were lying on the floor looking down or were concentrating on the robbers’ guns rather than their faces.

(ii) Defense Counsel’s Arguments

Counsel’s closing arguments spanned over 100 pages of trial transcript. Defendant’s attorney thoroughly detailed the potential unreliability of eyewitness identifications. She cautioned the jurors to pay special attention to the difficulty of identifying someone wearing a stocking mask, and pointed out inconsistencies and ambiguities in the eyewitnesses’ statements, both during and before trial.21 In addition, she emphasized that eight of the victims could not identify defendant. Counsel admonished the jury to remember that at the photo spread soon after the robbery—their first chance to identify the robbers—none of the victims positively identified defendant. She argued specifically that Erica Albertsen had seen defendant’s photograph at the initial spread and failed to recognize him. (Cf. People v. Walker (1986) 185 Cal.App.3d 155, 167 [229 Cal.Rptr. 591] [refusal to admit expert testimony on eyewitness identifications was harmless error, in part because defense counsel’s argument carefully reviewed the facts, “including the weakness of the eyewitness identification focusing on the psychological factors and unreliability of eyewitness identifications as they applied to (defendant.)”].)

Finally, defense counsel cited documented examples of convictions based on mistaken identifications, and described the results of a psychological study demonstrating that a person’s recollection of what he has seen may change over time. She emphasized that four months had passed between the *1149robbery and the trial, and that the witnesses’ recollections could be mistaken.

(iii) Jury Instructions Given

The court gave a variety of instructions relevant to eyewitness identifications. Although the giving of the general instructions may not alone preclude reversal for failure to give a specific eyewitness factor instruction (People v. West, supra, 139 Cal.App.3d at p. 610), their use is relevant, in combination with counsel’s arguments and cross-examination, to a determination of whether the error was prejudicial. (See People v. Coates, supra, 152 Cal.App.3d 665, 671.) In this case, the court’s instructions essentially covered each area of concern addressed in defendant’s requested instruction No. 3, albeit with less specificity than the requested instruction.

On the witnesses’ ability to observe the robbers, the court instructed the jurors at both the opening and close of trial that they were the “sole judges of the believability of the witnesses . . . and the weight to be given to the testimony of each,” and that in determining believability of a witness they could consider “the witness’ opportunity and ability to see and hear any matter about which he or she testifies and the witness’ ability to remember and relate those matters here in court . . . .” (CALJIC No. 2.20, as modified by the court, italics added.) In addition, before deliberations, the judge instructed that a witness’s observations may be unreliable: “It is a fact, also, that two people who witness an incident or a transaction often may see or hear it differently.” (CALJIC No. 2.21, italics added.) He further informed the jury that when determining the weight to be given to a witness’s opinion, “you should consider his or her credibility, the extent of the witness’ opportunity to perceive the matters upon which the opinion is based, and the reasons, if any, given for it.” (CALJIC No. 2.81, italics added.)

On the reliability of subsequent identifications, the court explained the prosecutor bore the burden of proving beyond a reasonable doubt that it was defendant who had committed the crimes in question. “You must be satisfied beyond a reasonable doubt of the accuracy of the identification of each defendant as the person who committed the offense before you may convict him. If, from the circumstances of the identification, you have a reasonable doubt whether a defendant was the person who committed the offense, then you must give the defendant the benefit of the doubt and find him not guilty.” (Former CALJIC No. 2.91, italics added.)

At another point, the court focused on the issue of prior inconsistent or erroneous identifications. It instructed the jury that in weighing the testimo*1150ny, it could consider “any statement previously made by a witness that is either consistent with any part of his or her testimony or any statement which is inconsistent with any part of the witness’ testimony.” (CALJIC No. 2.20, as modified by the court, italics added.)

Finally, in its instruction on the defense of alibi, the court referred to the possibility of misidentification, telling the jury: “If, after a consideration of all of the evidence, you have a reasonable doubt that a defendant was present at the time that the crime was committed, [the defendant] is entitled to an acquittal.” (CALJIC No. 4.50, italics added.)

Although these instructions did not list such specific factors as stress, lighting, and length of time available for the original observation, they directed the jury’s attention to the issue of the reliability of the identifications, and generally covered the areas on which defendant’s requested instruction focused. In addition, as described above, counsel’s extensive cross-examination and arguments further amplified the instructions given, and placed before the jury the factors specifically enumerated in the requested instruction.22

(c) No Indication of Jury Confusion

The presentation of evidence in this case took about eight days. The jury deliberated less than one day before finding defendant guilty.23 It did not manifest any confusion over defendant’s case; it never asked for clarification of instructions or rereading of transcripts, nor did it indicate any problem reaching a verdict as to defendant.

In contrast, the jury did demonstrate uncertainty concerning the guilt of the codefendant. It requested a rereading of testimony relating only to the codefendant, and, although finding defendant guilty on all counts, concluded it had reached an “impasse” with respect to the codefendant. Because the evidence against the codefendant, like that against defendant, was limited to eyewitness identification, the difference in outcome indicates that the jury thoughtfully considered all the evidence before it, and individually *1151evaluated the strength of the eyewitness evidence against each of the two defendants.24 In this respect, this case is similar to Walker, supra, 185 Cal.App.3d at page 166, in which the court’s rejection of expert testimony on eyewitness identifications was held harmless error based in part on the brevity of the jury deliberations (about one day after a six-day trial). The Walker court also found it significant that the jury acquitted the codefendant on one count, indicating the jury’s “concern with reaching a fair result.” (Ibid.)

The lack of jury confusion distinguishes this case from cases more closely balanced on the evidence. Compare West, supra, 139 Cal.App.3d 606, in which the jury asked for a rereading of a substantial portion of the testimony, and some instructions, and at one point reported itself deadlocked. (Id., at p. 610.) There the court noted that the jury’s requests for clarification demonstrated “the closeness of this case” and that therefore the error could not be deemed harmless. (Ibid.) In McDonald, supra, 37 Cal.3d 351, the jury deliberated 1910 hours after hearing evidence that took only 6 hours to present. We stated that when jurors “deliberate in these circumstances for more than three and a half times longer than it took to put on the entire prosecution and defense cases, we may fairly infer they found the issue difficult to decide.” (Id., at p. 376, fn. 23; cf. Walker, supra, 185 Cal.App.3d at p. 166.) Similarly, in People v. Coates, supra, 152 Cal.App.3d 665, the court, in finding the failure to instruct on eyewitness identification factors prejudicial, found it significant that the jury sought a rereading of evidence and the reasonable doubt instruction. The court stated, “[It] is as though the jurors themselves were requesting a more detailed instruction on the subject of identification, the only issue in the case.” (Id., at p. 672.) No such inference of jury confusion can be drawn here.

In sum, the record reveals that the factors listed in defendant’s requested instruction were adequately presented to the jury at trial, and that accordingly the failure to give requested instruction No. 3 was harmless error. In the course of questioning the witnesses who identified defendant, defense counsel elicited testimony on the difficulty of seeing the robbers through their stocking masks, the fear and panic the witnesses felt during the robbery, and their initial uncertainty about the identification of defendant. In argument, both defense attorneys stressed the unreliability of eyewitness *1152identifications, and specifically emphasized the factors relevant to the reliability of the eyewitness identifications and supported by the evidence. Further, the trial court instructed the jury on the credibility of witnesses, the reliability of witnesses’ opinions, and the burden of proof, which, in combination with the specific factors brought out by defense counsel, thoroughly informed the jury of the issues involved in its determination of defendant’s guilt. Finally, the jury distinguished between defendant and codefendant, demonstrating that it carefully assessed the strength of the identification evidence against each of them. It seems apparent that as to defendant the jury found the eyewitness testimony credible, even in light of all the factors that were brought out at trial, and that it found defendant’s alibi unpersuasive. We hold it is not reasonably probable that a verdict more favorable to defendant would have resulted had the court given defendant’s requested instruction No. 3.

D. Defendant’s Proposed Cautionary Instruction

Defendant’s proposed instruction No. 5 would advise the jury that eyewitness identification testimony may be mistaken and “should be received with caution.”25 We hold this instruction was properly refused.

No California case has held that such a cautionary instruction is required in addition to the eyewitness “factors” instruction. As stated in People v. McCowan, supra, 85 Cal.App.3d at page 679, an instruction stating that mistaken identification is “not uncommon” is objectionable and may be properly rejected because it “does not state a principle of law or establish a basis for instructing a jury.” (But see Martinez, supra, 191 Cal.App.3d 1372, 1381 [“Assuming, arguendo,” it was error to refuse to admonish the jury to consider with caution eyewitness identification testimony, any error was harmless because the eyewitnesses were “thoroughly cross-examined” and the jury was given general instructions including CALJIC Nos. 2.20 and 2.92].)

We are aware of no jurisdiction in which a cautionary instruction such as defendant’s No. 5 has been deemed appropriate. In fact, at least one other court that considered such an instruction expressly rejected it. In State v. Watson (W.Va. 1984) 318 S.E.2d 603, the defendant was charged with breaking and entering. The only testimony placing him in the burglarized home was that of an 11-year-old child. Although the court held that the *1153defendant was entitled to a specific instruction on identification, it refused a request for a special cautionary instruction similar to that requested here. (Id., at pp. 614-615.) The West Virginia Supreme Court of Appeals reasoned in part that the proposed cautionary instruction was essentially premised on “the theory that any eyewitness testimony is suspect, [which is] a legal point that we believe is too broad.”26 (Id., at p. 615.)

Defendant’s requested instruction No. 5 is both improper and unnecessary. A special cautionary instruction is unnecessary because the “factors” instruction already required properly highlights the factors relevant to defendant’s concerns about the reliability of eyewitness identification testimony in a particular set of circumstances.

The additional general cautionary instruction requested here would produce redundancies which could unbalance the jury’s deliberative process. Singling out a category of evidence for special consideration may cause the jury to give it undue weight in its deliberations. The cautionary instruction given along with the factors instruction would place unwarranted emphasis on the eyewitness identification, and likely give the jury the improper impression that the court considers the eyewitness identification evidence not only particularly important but also overwhelmingly suspect.

The requested cautionary instruction would also improperly usurp the jury’s role as the exclusive trier of fact by binding it to the view that eyewitness identifications are often mistaken. It goes beyond previously approved cautionary instructions by informing the jury not only that it should view the eyewitness identification evidence with caution, but also that such evidence is frequently unreliable and misidentification is “not uncommon.”27 This is not an appropriate function of an instruction. If the *1154defendant wishes to present to the jury information on the unreliability of eyewitness identifications under a particular set of circumstances, he must use means other than a jury instruction, such as expert testimony. (See McDonald, supra, 37 Cal.3d 351.) As we stated ante, expert testimony has the advantage of being subject to cross-examination and rebuttal, thus allowing the jury to determine for itself the weight it should give to expert opinions, rather than binding the jury to accept certain experts’ views.

In sum, the eyewitness “factors” instruction provides the jury with sufficient means to evaluate eyewitness identification testimony and alerts jurors to the factors that may affect eyewitness identifications. In addition, expert testimony may be used when appropriate to further elucidate the effect of the factors listed. Accordingly, we reject the use of a special additional and cumulative cautionary instruction regarding the unreliability of eyewitness identification. The trial court did not err in refusing to give this requested instruction.

III. Disposition

For the reasons discussed above, we hold that the trial court did not commit reversible error. The judgment is affirmed in its entirety.

Panelli, J., Arguelles, J., Eagleson, J., and Kaufman, J., concurred.

MOSK, J.

I dissent.

I agree with the majority that the trial court erred in refusing to give an instruction focusing the jury’s attention on psychological factors that could have adversely affected the accuracy of the eyewitness identifications and hence could have given rise to a reasonable doubt of defendant’s guilt. Indeed, we said as much in People v. McDonald (1984) 37 Cal.3d 351, 377 [208 Cal.Rptr. 236, 690 P.2d 709, 46 A.L.R.4th 1011], footnote 24, and the cases there cited. But I strongly disagree with the majority’s further conclu*1155sions (1) that the wording of the particular charge refused in this case—or of CALJIC No. 2.92—would have been adequate to satisfy defendant’s right to such an instruction, and (2) that the error in refusing to give a correct instruction on this topic was harmless.

I

Defendant’s requested instruction No. 3 purported to list the factors in the record that could have affected the accuracy of the eyewitness identifications.1 The majority hold this instruction was adequate to the task, and that the court erred in refusing to give it. The instruction thus approved by the majority, however, was deficient in two respects.

First, it omitted a number of other relevant factors shown by the evidence. Perhaps most significant, the approved instruction was silent on the “cross-racial factor” that often operates when the witness and the offender are of different races.2 In McDonald we explained the importance of this factor in detail. (37 Cal.3d at p. 368.) When, as here, a defendant’s liberty turns wholly on cross-racial identifications, we should not approve a “factor” instruction that fails even to mention the cross-racial effect.

Other relevant factors are also missing from the instruction approved by the majority. Thus it did not discuss the familiarity effect, i.e., that an identification of a person whom the witness has had a prior opportunity to observe is more reliable than an identification of a stranger. Yet both kinds *1156of identification were involved here: Janice Tong, who identified codefendant Wellington, had seen and spoken with him before the robbery, while none of the three witnesses who identified defendant had ever seen him before.

Nor did the approved instruction mention other factors we recognized in McDonald (37 Cal.3d at p. 368), e.g., that a witness may subconsciously incorporate into his memory post-event information—both correct and mistaken—acquired from other sources, such as descriptions by other witnesses or press reports of the event (see Loftus, Eyewitness Testimony (1979) pp. 54-87), or that the accuracy of the witness’s recall may be adversely affected by unconscious biases or cues in identification procedures and methods of questioning (id. at pp. 89-99).

As the majority acknowledge, we held in the leading case of People v. Hall (1980) 28 Cal.3d 143, 159 [167 Cal.Rptr. 844, 616 P.2d 826], that when a defendant is entitled to an instruction on eyewitness identification factors, it is error for the court to refuse to cure defects in the charge requested—for example, that it is overinclusive or underinclusive—and simply give no such instruction at all. In Hall the requested charge was defective because it included factors not shown by the evidence; here, on the contrary, it was defective because it omitted factors that the record did support. Yet the duty of the trial court remained the same: People v. Coates (1984) 152 Cal.App.3d 665, 670-671 [199 Cal.Rptr. 675], held that the proffered instruction was defective because it failed to include an important factor shown by the evidence, “But the court, as Hall states, should have tailored the instruction by inserting the parts it believed should not have been omitted.” (Id. at p. 671.) The court’s failure to do so, Coates concluded, was reversible error.

Here defense counsel respectfully referred the trial court to the foregoing rule of case law, and begged the court to point out any perceived defects in the proposed instruction that could be cured by amendment; but the court refused to do so, erroneously denying that it had any responsibility in the matter.3

*1157The instruction approved by the majority was also deficient in a more fundamental respect: it merely listed the factors that it told the jurors they “should consider,” failing to inform them what effect any such factor may have on the accuracy of the eyewitness identifications in this case. The same defect appears in CALJIC No. 2.92, which the majority also approve (and quote in appendix): although it refers to more factors than defendant’s requested instruction No. 3, it too merely lists them with no explanation of their effect. And the majority expressly endorse this format: a proper instruction on eyewitness identification factors, say the majority, deals with the issue simply “by listing, in a neutral manner, the relevant factors supported by the evidence.” (Ante, p. 1141, italics added.)

I do not contend that every factor in the instruction must be explained to the jury: we may assume that from personal experience jurors know how an eyewitness identification can be adversely affected by the common circumstance, for example, that the lighting was dim or the witness was at some distance from the scene—or, as here, that the opportunity to observe was severely limited and the perpetrator was masked. Nor do I contend, contrary to the majority’s distortion of my views, that the jury should be instructed on any factor that is “a matter of ongoing scientific debate.” (Ante, p. 1143.) Indeed, I explain below that the CALJIC instruction approved by the majority is defective precisely because, inter alia, it encourages the jury to speculate on the highly debatable issue of the effect of “stress” on an eyewitness identification. And I there conclude, as the majority urge, that “Such an instruction should not take sides in an ongoing scientific controversy.” (Post, p. 1160.)

Nevertheless, there are a number of other factors—including some that are often crucial to the accuracy of the identification—that are not in dispute among behavioral scientists but may not be familiar to jurors. In addition to the foregoing more obvious circumstances, “other factors bearing on eyewitness identification may be known only to some jurors, or may be imperfectly understood by many, or may be contrary to the intuitive beliefs of most.” (People v. McDonald, supra, 37 Cal.3d at p. 368.) Yet the same factors have been extensively studied by behavioral and social scientists, who have reported their methods, data, and conclusions in numerous books and articles published in the professional literature. (Id. at pp. 364-365.) After reviewing that literature in McDonald we concluded, “The consistency of the results of these studies is impressive, and the courts can *1158no longer remain oblivious to their implications for the administration of justice.” (Id. at p. 365.)

The majority appear to deny there is any such consensus, but their denial is refuted by a reading of the treatises and articles we cite in McDonald (ibid.). The majority then claim that an instruction explaining the effect of unfamiliar but noncontroversial eyewitness identification factors “would of necessity adopt the views of certain experts and incorporate the results of certain psychological studies while discounting others.” (Ante, p. 1141, italics added.) But as to a number of important factors there simply are no significant “other” views or studies to “discount.” For example, I challenge the majority to cite a single reputable professional study that concludes there is no such thing as the cross-racial effect. (See Johnson, Cross-Racial Identification Errors in Criminal Cases (1984) 69 Cornell L.Rev. 934, 938-939 [citing 10 studies documenting the cross-racial effect].)

It was in large part because of this consistency in the results of current eyewitness identification studies that we held in McDonald that when such identification is both crucial and uncorroborated the defendant is entitled to introduce expert testimony on the factors shown by the record that could have affected the identification. (37 Cal.3d at p. 377.) It is no great leap to now hold that in the absence of that testimony the jurors should be given an instruction, if requested, that not only tells them the bare name of each factor but also gives them some understanding of the specific effect of factors that are noncontroversial but nevertheless are not generally understood by laypersons. “If the instructions convey no data, telling jurors to ‘consider’ a factor they know nothing about may only confuse them.” (Johnson, op. cit. supra, 69 Cornell L.Rev. at p. 985.) Left to their own devices, the jurors will inevitably fall back on such sources as folk wisdom, private beliefs, or sheer speculation. Unfortunately, the effect of certain important psychological factors may be the exact opposite of what the average juror expects.

The defects in CALJIC No. 2.92 illustrate the risks all too well. For example, the instruction tells the jurors to consider “The cross-racial or ethnic nature of the identification,” but it gives them no further guidance on the topic. Yet there is more to be said. As we explained in McDonald (37 Cal.3d at p. 368), “To be sure, many jurors are likely to have some awareness of the fact that an eyewitness is more accurate in identifying a person of his own race than one of another race. [Citation.] But it appears that few jurors realize the pervasive and even paradoxical nature of this ‘own-race effect,’ information that has emerged from numerous empirical studies of the question. These studies establish that the effect is strongest when white witnesses attempt to recognize black subjects; in such circumstances ‘The *1159impairment in ability to recognize black faces is substantial.’ ([Johnson, op. cit. supra, 69 Cornell L.Rev. at pp. 938-939.]) . . . The studies also reveal two aspects of the matter that will probably be contrary to most jurors’ intuitions: first, that white witnesses who are not racially prejudiced are just as likely to be mistaken in making a cross-racial identification as those who are prejudiced; and second, that white witnesses who have had considerable social contact with blacks may be no better at identifying them than those who have not. (Id. at pp. 943-944.) Finally, some jurors may deny the existence of the own-race effect in the misguided belief that it is merely a racist myth exemplified by the derogatory remark, ‘they all look alike to me,’ while others may believe in the reality of this effect but be reluctant to discuss it in deliberations for fear of being seen as bigots. (Id. at p. 969; see also Wells. A Reanalysis of the Expert Testimony Issue, in Eyewitness Testimony: Psychological Perspectives [Wells & Loftus, edits. (1984)], p. 309.)” (Fn. omitted.) The CALJIC instruction approved by the majority conveys none of this information whatever.

The same instruction also tells the jury, again without explanation, to consider “The extent to which the witness is either certain or uncertain of the identification.” This portion of the CALJIC charge is even more misleading. The average juror doubtless takes it as confirming the widespread lay belief that the more certain an eyewitness is of his identification, the more likely the identification is correct. Yet that belief is apparently mistaken: as we explained in McDonald (37 Cal.3d at p. 369), there is in fact a “lack of correlation between the degree of confidence an eyewitness expresses in his identification and the accuracy of that identification. Numerous investigations of this phenomenon have been conducted: the majority of recent studies have found no statistically significant correlation between confidence and accuracy, and in a number of instances the correlation is negative—i.e., the more certain the witness, the more likely he is mistaken. (Wells & Murray, Eyewitness Confidence, in Eyewitness Testimony: Psychological Perspectives, pp. 159-162.) Indeed, the closer a study comes to reproducing the circumstances of an actual criminal investigation, the lower is that correlation (id. at pp. 162-165), leading the cited authors to conclude that ‘the eyewitness accuracy-confidence relationship is weak under good laboratory conditions and functionally useless in forensically representative settings.’ (Id. at p. 165; see also Deffenbacher, Eyewitness Accuracy and Confidence: Can We Infer Anything about their Relationship? (1980) 4 Law & Human Behav. 243.) The average juror, however, remains unaware of these findings: ‘A number of researchers using a variety of methods have found that people intuitively believe that eyewitness confidence is a valid predictor of eyewitness accuracy.’ (Wells & Murray, supra, at p. 159, citing five recent studies.)” Thus rather than correcting this common misconception of jurors, the CALJIC instruction actually reinforces it.

*1160In addition to being underinclusive, CALJIC No. 2.92 manages to be overinclusive. It directs the jury to consider “The stress, if any, to which the witness was subjected at the time of the observation.” Again the jury is left to speculate on what effect “stress” is supposed to have on an eyewitness who is also a victim. On this question, however, there is no consensus among laymen. Some jurors may assume that the more stress an eyewitness is under, the more likely his identification is correct—believing that the stress focuses the witness’s attention on the details of the scene and particularly on the identity of his assailant. The view is typified by victims who remark, “I’ll never forget that face.” But other jurors may assume the exact opposite, i.e., that greater stress means poorer identification, believing that an eyewitness in fear of life or limb focuses his attention instead on how serious is the threat against him and how he can escape from it. This view is typified by victims who remark, “I couldn’t take my eyes off his gun.” Indeed, several eyewitnesses in the case at bar testified to precisely that effect.

Not only do laymen not agree on the effect of stress on eyewitness identification—behavioral scientists do not agree either. A recent review of the relevant literature disclosed nine studies suggesting that stress improves (or at least does not worsen) eyewitness accuracy, and ten suggesting the contrary. (Egeth & McCloskey, Expert Testimony about Eyewitness Behavior: Is it Safe and Effective? in Eyewitness Testimony: Psychological Perspectives, supra, p. 297.) Some researchers believe that both effects occur, depending on the level of stress: under this view—termed the “Yerkes-Dodson law”— mild stress increases eyewitness accuracy while severe stress decreases it. (See, e.g., Loftus, Eyewitness Testimony, supra, pp. 33-36.) But even if the “Yerkes-Dodson law” were as undisputed as the law of gravity, few if any jurors know of it. And it is far from undisputed: other researchers find no empirical evidence to support this “law” (Egeth & McCloskey, op. cit. supra, at p. 297), and question whether it applies outside the laboratory (Yarmey, The Psychology of Eyewitness Testimony (1979) p. 52).

The fact that the experts disagree on the existence and effect of a “stress factor” does not mean, of course, that it should not be the subject of expert testimony (McDonald, 37 Cal.3d at p. 369, fn. 15); but it does mean that this factor should be omitted from any instruction—like CALJIC No. 2.92— that advises juries of the factors they may consider in the absence of such testimony. Such an instruction should not take sides in an ongoing scientific controversy.4

*1161II

I further disagree with the majority’s conclusion that the error in refusing to give a correct instruction on the factors affecting the eyewitness identifications was harmless on the record of this case. The majority offer three reasons why the error is allegedly not reversible, but none is convincing.

A.

The majority begin by asserting that the evidence was sufficient to support the verdict, and stress that defendant does not contend it was insufficient. (Ante, p. 1145 & fn. 17.) This is strange reasoning indeed. It implies that after a trial in which there was instructional error the judgment will be affirmed on appeal unless the defendant also contends—and a reviewing court also holds—that the evidence of guilt is insufficient as a matter of law. Such a rule would obviously be both unprecedented and indefensible.

Perhaps the majority is simply attempting to portray the prosecution case in the best possible light. Yet the effort founders on the reality of the trial. The majority emphasize that “three of the eleven victims identified defendant as one of the robbers.” (Id. at p. 1145.) Of course, this means that eight of the eleven victims could not identify him, and the fact that more than three-quarters of the eyewitnesses were unable to identify defendant underscores the weakness of the prosecution case. The majority seek to explain away this weakness by asserting that the latter eight witnesses were “for the most part unable to describe the faces of any of the robbers.” (Ibid., italics in original.) But why were they unable to do so? Not because of any particularity of these eight witnesses, but because the circumstances of the crime made an identification by any witness extremely difficult. Yet these circumstances affected the three witnesses who purported to identify defendant no less than the eight who could not.

The majority also reason that “Although there was no evidence corroborating the eyewitness identifications, neither did any eyewitness testify at trial that defendant was not one of the robbers.” (Ibid., italics added and *1162deleted.) This is a non sequitur: the fact that no witness said defendant was not among the robbers does not detract in the slightest from the fact that the only evidence connecting him with the crime was the uncorroborated eyewitness testimony. The two facts are as unrelated as the proverbial apples and oranges.

The majority also exaggerate the importance of the fact that no witness testified defendant was not among the robbers. The majority note that such testimony appeared in McDonald (37 Cal.3d at pp. 358-359) and in People v. West (1983) 139 Cal.App.3d 606, 608 [189 Cal.Rptr. 36], and jump to the conclusion that “On this basis” the case at bar is distinguishable from McDonald and West. But a fair reading of the latter opinions shows they cannot be dismissed so easily: in McDonald we list the exculpatory testimony as only one circumstance among several that support our conclusion of prejudice (37 Cal.3d at p. 376), and most of the other circumstances—e.g., the absence of any other evidence connecting the defendant with the crime, the weaknesses in the prosecution eyewitness testimony—are also present in the case at bar. The same is true of West (139 Cal.App.3d at p. 610). And the majority ignore other decisions closely similar to McDonald and West in which there was no testimony whatever that the defendant was not the criminal, yet which nevertheless found prejudice. (People v. Coates, supra, 152 Cal.App.3d 665, 671-672; People v. Palmer, supra, 154 Cal.App.3d 79, 89.) Clearly such exculpatory testimony is not a prerequisite to prejudice.

Finally, the majority assert that in their opinion the defense put on by defendant was “not strong” and could have been “discounted” by the jury. (Ante, p. 1145.) The majority seem concerned that the defense was alibi and the two witnesses who presented it were related to defendant; but likewise in McDonald the defense was alibi, and four of the six witnesses who presented it were also related to the defendant. On the basis of the cold record the defense in the case at bar is no less credible than the McDonald alibi, and it is for the jury rather than this court to decide whether to believe it.

B.

The second justification offered by the majority for their conclusion that the error here was harmless is that the factors omitted from the instructions were nevertheless “put before the jury” by cross-examination, closing arguments, and other instructions. (Ante, p. 1146.) Again reality is otherwise. As explained in a major review of this issue, “Cross-examination and closing arguments of counsel . . . inadequately protect criminal defendants from misidentification in cases in which eyewitness identification testimony is important. Concededly, counsel may successfully expose and emphasize factors suggesting that a particular eyewitness’ identification is unreliable. *1163Jurors, however, are aware that the defense attorney, as adversary, is obligated to defend his client vigorously. For this reason, the jury is likely to discount any attacks the defense makes on the reliability of an identification as being motivated by his partisan desire to win. In addition, the generally acknowledged natural inclination of jurors to afford great credence to eyewitness identifications also diminishes the effectiveness of cross-examination and closing arguments in minimizing the risks of misidentification. Because the trial judge is a neutral officer of the court, a jury is likely to be more receptive to his nonpartisan instructions which enumerate the factors that tend to render an eyewitness identification unreliable.” (Fns. omitted.) (Note, Eyewitness Identification Testimony and the Need for Cautionary Jury Instructions in Criminal Cases (1983) 60 Wash.U.L.Q. 1387, 1422-1423.)

Nor do cross-examination and argument miraculously become adequate for this purpose, as the majority claim, when they are viewed in combination with general instructions on how the jury should weigh the testimony of any witness (CALJIC Nos. 2.20, 2.21, and 2.81), on proof beyond a reasonable doubt (CALJIC No. 2.91), and on alibi (CALJIC No. 4.50). It was precisely the insufficiency of such general instructions in this context that led our courts to hold in People v. Hall, supra, 28 Cal.3d 143, 159-160, and its progeny, that a defendant is entitled to a specific instruction on psychological factors affecting eyewitness identifications. When insufficient general instructions are added to ineffective remarks of counsel, the result is still inadequacy: zero plus zero equals zero.

C.

The majority’s final argument against prejudice is also their weakest: they rely on the fact that the jury did not ask that any testimony be reread or any instruction be clarified, and deliberated less than one day before finding defendant guilty. It is true that in McDonald (37 Cal.3d at p. 376, fn. 23) the jury took longer to convict, and that in People v. West, supra, 139 Cal.App.3d 606, 610, and People v. Coates, supra, 152 Cal.App.3d 665, 672, the jury asked for a rereading of certain testimony and instructions. But in each case there was ample other evidence pointing to prejudice, and the procedural facts now relied on by the majority were mentioned largely as makeweights; indeed, in McDonald we relegated the entire discussion to a footnote. Again, therefore, the majority overstate the point. As with their observation that defendant does not claim insufficiency of the evidence, so here: the majority imply that a judgment will be affirmed on appeal despite instructional error unless the jury asked for a rereading of testimony or instructions or deliberated for longer than an unspecified minimum time. *1164Surely this would be an unprecedented and indefensible rule of appellate review.

The majority also stress that the jury did ask for a rereading of the testimony relating to codefendant Wellington, and was unable to reach a verdict as to him; the majority assert that the evidence against Wellington was also “limited to eyewitness identification,” and conclude that the difference in outcome means the jury “thoughtfully considered” all the identification evidence against both defendants. (Ante, p. 1150.)

Here lies the ultimate irony in this troubling case. Although the evidence against Wellington was eyewitness identification testimony, it was stronger on several grounds than the eyewitness identification testimony against defendant. Wellington was identified at trial by Janice Tong, wife of a police officer. First, as opposed to most of the other witnesses Tong had ample opportunity to observe Wellington during the robbery. From her position she had a clear view of him, and testified that she watched him for “90 percent of the time” that the robbers were in the office, seeing his face from several angles. On three occasions he heard her move, looked directly at her, and ordered her to keep still. Later he came over to her, took her purse and emptied it on the floor, pocketing some of its contents. She explained to the jury that she could make out Wellington’s features because his stocking mask was not tight-fitting and only his nose was slightly flattened; she also described his clothes in detail.

Second, following the crime Tong gave the investigating officers a description of Wellington, then selected his picture from a photographic array and picked him out of a lineup. She concluded by testifying that she had “no doubt” whatever that Wellington was one of the robbers.

Third and most important, Tong was the only witness who had previously seen one of the robbers when he was not wearing a mask. She testified that about one week before the robbery Wellington appeared unannounced in an area of the office reserved for employees; she accosted him and inquired if he needed help; he asked her for a job application, and left. He was, moreover, wearing the same clothes that he wore in the robbery.

None of the foregoing was true as to defendant—yet defendant was convicted while Wellington was not. Contrary to the majority’s conclusion, this difference in outcome strongly suggests that the jury did not “thoughtfully consider” the eyewitness identification testimony, and that it needed instructional assistance from the court in doing so.

We declared in McDonald, 37 Cal.3d at page 376, that “An error that impairs the jury’s determination of an issue that is both critical and closely *1165balanced will rarely be harmless.” This is such an error. Here, as in West, supra, 139 Cal.App.3d 606, Coates, supra, 152 Cal.App.3d 665, Palmer, supra, 154 Cal.App.3d 79, and McDonald, supra, 37 Cal.3d 351, identity was the sole issue at trial and the only evidence connecting defendant with the crime was the eyewitness identification testimony. Also as in those cases, the record discloses a number of factors that could have raised reasonable doubts as to the accuracy of the identifications: the events were sudden and unexpected; defendant was a stranger to the witnesses; their opportunity to observe was brief and discontinuous; their view of the robbers’ faces was seriously impaired by the fact that all wore stocking masks and took pains to prevent the witnesses from looking at them; most of the witnesses were unable to identify defendant either after the crime or at trial; the three witnesses who did identify him were each of a different race from defendant; they also admitted they were afraid during the encounter, they had varying degrees of difficulty in identifying defendant from photographs or in lineups, and in fact they made several misidentifications.

In these circumstances it is reasonably probable that the admitted error in the eyewitness identification instructions contributed to the verdict and resulted in a miscarriage of justice. Defendant is entitled to a new trial on correct instructions.

Broussard, J., concurred.

Appendix

CALJIC NO. 2.92 Factors to Consider in Proving Identity by Eyewitness Testimony

Eyewitness testimony has been received in this trial for the purpose of identifying the defendant as the perpetrator of the crime[s] charged. In determining the weight to be given eyewitness identification testimony, you should consider the believability of the eyewitness as *1166well as other factors which bear upon the accuracy of the witness’ identification of the defendant, including, but not limited to, any of the following:

[The opportunity of the witness to observe the alleged criminal act and the perpetrator of the act;]

[The stress, if any, to which the witness was subjected at the time of the observation;]

[The witness’ ability, following the observation, to provide a description of the perpetrator of the act;]

[The extent to which the defendant either fits or does not fit the description of the perpetrator previously given by the witness;]

[The cross-racial or ethnic nature of the identification;]

[The witness’ capacity to make an identification;]

[Evidence relating to the witness’ ability to identify other alleged perpetrators of the criminal act;] I

[Whether the witness was able to identify the alleged perpetrator in a photographic or physical lineup;]

[The period of time between the alleged criminal act and the witness’ identification;] [Whether the witness had prior contacts with the alleged perpetrator;]

[The extent to which the witness is either certain or uncertain of the identification;] [Whether the witness’ identification is in fact the product of his own recollection;]

Any other evidence relating to the witness’ ability to make an identification.