Cite as: The People &c. v. Edwin Santiago, Index 166, NYLJ 1202519665792, at *1 (Ct. of App., Decided October 20, 2011)
Opinion by Judge Pigott. Chief Judge Lippmanand Judges Ciparick, Graffeo, Read, Smith and Jones concur.
Decided October 20, 2011
Jeffrey I. Dellheim, for appellant.
Patrick J. Hynes, for respondent.
*1EUGENE F. PIGOTT, JR. J.:
In this case, turning on the accuracy of eyewitnesses' recognition of an assailant's partially concealed face, we consider whether two additional eyewitness identifications sufficiently corroborated the victim's identification of the defendant, so as to render expert testimony on eyewitness*2recognition memory unnecessary. We conclude that they did not, and that it was error to exclude much of the proposed testimony.
In the early morning of January 10, 2003, a woman waiting for a train at a Manhattan subway station was attacked by a stranger. She had noticed the man when, after they made eye contact on the platform, he stepped behind a pillar before approaching her. When he was an arm's length away from her, the man asked whether she was "working." When she inquired what he meant, he asked if she was an escort. After she looked away and said "No," the man began assaulting her. She closed her eyes while raising her hands to protect herself. She could not tell whether her assailant had a weapon. After about ten seconds, the attack stopped, and the assailant fled.
The victim was assisted by workers at the station, who, along with the police, searched fruitlessly for the assailant. An ambulance transported her to a hospital where she was treated. At the hospital, the victim gave detectives a description of her attacker, a Hispanic male, late 20s or early 30s, five feet, eight inches to five feet, nine inches tall, with a mustache and a goatee. To one detective, the victim described her assailant's "brown/yellow mustache."
The man was wearing a winter jacket, a hooded sweatshirt or "hoodie," jeans, and a "winter hat." The jacket, hoodie and hat together covered the assailant's head in such a *3way that his face was concealed "from the middle of his top lip, down, and from the top of his eyebrows up." The victim could not see her assailant's hair, except for his eyebrows and mustache. On the day after the attack, the victim was interviewed by a police artist who created a sketch of the perpetrator.
Police detectives visited the subway station in search of eyewitnesses. Edwin Rios, who had seen the assault, described the assailant as a Hispanic male in his mid- to late-20s, with a goatee, wearing a hood. The assailant had passed Rios after the attack, carrying a knife. Later, police located another witness, Pablo Alarcon, who had noticed the assailant beforehand because the man's facial expression made Alarcon nervous. After the attack, Alarcon saw the assailant put a knife away as he fled. Alarcon also described the perpetrator as a Hispanic male with a goatee. Police officers showed Rios a copy of the artist's sketch of the perpetrator; Rios thought it was "more or less" accurate.
On January 19, 2003, a plain-clothes police officer patrolling a subway station in Brooklyn noticed a man in a winter jacket, jeans and winter cap selling MetroCard "swipes." Later, the officer saw the same man engaged in the same activity at the next stop. The officer arrested the man, defendant Edwin Santiago, and his photograph made its way to the detective squad investigating the January 10 attack.
An array comprising photographs of six men and *4including defendant's arrest photograph was shown to Alarcon on January 22. He claimed not to recognize defendant or any of the other men in the array. The victim viewed the photographic array on January 24. She testified that, when she saw the photograph of Santiago, it felt as if her "heart stopped and [she] got really scared and  said that that was him."
*5Santiago, who had been released, was rearrested the following day, at a shelter for the homeless. A photograph taken after the second arrest shows Santiago with a dark mustache and goatee. He was 30 years old and five feet, five inches in height.
On January 26, the victim identified Santiago in a six-person line-up. According to the victim, when she saw defendant, she felt "really scared"—it was, once again, as if her "heart stopped" and she "knew it was him." On the same day, Alarcon viewed the line-up. As he later explained at defendant's suppression hearing and trial, he recognized defendant as the perpetrator of the attack with an "eighty percent" feeling of confidence in his identification, but, because he was concerned about his immigration status, he told the police that he did not recognize anyone. The following day, Alarcon saw a photograph of Santiago in handcuffs, accompanied by police officers, in a Spanish-language newspaper; the article made it clear that Santiago had been identified by the victim of the subway attack and arrested.
Santiago was indicted by a grand jury on a first-degree assault charge on February 6, 2003.
No physical evidence linked defendant to the assault. As it appeared at the time of the indictment, the People's case would be built entirely on the victim's identification. Attempting to secure expert testimony from Professor Steven Penrod on the psychological factors affecting the accuracy of eyewitness identification, Santiago filed a motion in limine on June 26. The People countered by arguing, among other things, that much of the proposed expert testimony was within the common understanding of jurors, or inapplicable to the Facts of the case.
Supreme Court granted defendant's motion to the extent of ordering a Frye hearing (see Frye v. United States, 293 F 1013 [DC Cir 1923]), to determine whether the principles Professor Penrod proposed to describe in his testimony had gained general acceptance in their scientific fields. During subsequent motion practice, defendant gave provisional summaries of Penrod's expected testimony. He would testify concerning studies that support various principles proposed by psychologists in the field of eyewitness recognition—exposure time (the amount of time available for viewing a perpetrator affects the witness's ability to identify the perpetrator); cross-racial and cross-ethnic inaccuracy (non-Hispanic Caucasian eyewitnesses are generally *6less accurate in identifying Hispanic people than in identifying other non-Hispanic Caucasians); weapon focus (a victim's focus on the weapon used in an assault can affect ability to observe and remember the attacker); lineup fairness (similarity of fillers to the suspect increases identification accuracy); lineup instructions (police instructions indicating that the police believe the perpetrator to be in the lineup increase the likelihood of false identification); forgetting curve (the rate of memory loss for an event is greatest right after the event and then levels off over time); postevent information (eyewitness testimony about an event often reflects not only what the witness actually saw but also information the witness obtained later); wording of questions (eyewitness testimony about an event can be affected by how questions put to the witness during investigation are worded); unconscious transference (eyewitnesses sometimes identify as the culprit an individual familiar to them from other situations or contexts); simultaneous versus sequential lineups (witnesses are more likely to make mistakes when they view simultaneous lineups than when they view sequential lineups); eyewitness confidence issues (an eyewitness's confidence level is not a good predictor of eyewitness accuracy, but eyewitness confidence is the major determinant in whether an identification is believed by jurors), and confidence malleability (eyewitnesses' confidence levels can be influenced by factors unrelated to identification accuracy).
*7After defendant's case was transferred to a different Supreme Court Justice, defendant renewed his motion to admit expert testimony. On December 19, 2003, Supreme Court denied defendant's motion, without holding a Frye hearing. While noting that there was "no corroboration of the victim's identification," Supreme Court nonetheless ruled that the case was not "an appropriate one for an expert identification witness" (2 Misc 3d 652, 653 [Sup Ct 2003]). Supreme Court found that the case did not involve a cross-racial identification as contemplated by the psychological literature, and that weapon focus was not relevant because the victim was not aware of any weapon. Supreme Court further reasoned that testimony about how lineup instructions can influence an identification would be inappropriate, because the victim here "must have realized that the person whose photograph she selected would be in the lineup" (id. at 654). The court rejected expert testimony on postevent information, the forgetting curve, the wording of questions, and eyewitness confidence issues (including confidence malleability), on the ground that "by helping to create the sketch and approving its final version, the victim went on the record, for better or worse, about the facial features of her attacker long before any of those topics potentially could influence her lineup identification" (id. at 654). The court found evidence about simultaneous versus sequential lineups inappropriate on the basis that "[j]urors are not experts on constitutional law and *8procedure, and cannot be educated about those topics during a trial" (id. at 655). In addition, Supreme Court noted that, in Frye hearings in other cases, postevent information and unconscious transference had been found to be not generally accepted within the relevant scientific community.1 Supreme Court concluded by inviting the parties to "suggest ways to address, during jury selection and in the final instructions, the topic of a witness's confidence if one side feels that the court's charge on 'certainty' is not adequate to cover 'confidence'" (id. at 655). Defense counsel objected to the ruling.
On December 21, 2003, Alarcon told an assistant district attorney that he had in fact recognized someone in the January 2003 lineup. Shown photographs of that lineup, he identified defendant as the person he had recognized from the attack. On January 20, 2004, Rios was shown a new lineup, in which he identified Santiago.
Following a suppression hearing, at which his motions were denied, Santiago was tried before a jury. The victim identified Santiago at the trial, insisting that there was no doubt in her mind that he was her assailant. Rios and Alarcon also identified defendant; the latter repeated that he was only *9"eighty percent sure" of his identification. The jury heard about the prior photographic and lineup identifications.
After the defense rested, Santiago renewed his objection to Supreme Court's refusal to admit the expert testimony, moving to reopen his case. Supreme Court denied the motion.
Defense counsel's summation stressed that certainty of identification was not equivalent to accuracy of identification. As promised, Supreme Court's final charge told the jury to "[k]eep in mind that the witness's confidence or lack of confidence in his or her testimony is not necessarily indicative of accuracy of identification."
On February 5, 2004, the jury found Santiago guilty of assault in the first degree. Supreme Court convicted him accordingly, sentencing him to 25 years imprisonment, to be followed by 5 years postrelease supervision.
Santiago timely appealed, raising several issues he has since abandoned, as well as the argument we now consider—his claim that Supreme Court abused its discretion in denying his request for expert testimony on eyewitness identification. On May 6, 2010, a fractured Appellate Division affirmed Supreme Court's judgment. Two Justices concluded that there was no abuse of discretion; another concurred, in an opinion expressing the view that any error was harmless. The two dissenting Justices would have held that Supreme Court abused its discretion, and one *10of the dissenting Justices granted defendant leave to appeal to this Court. We now reverse.
A trial court may, in its discretion, admit, limit, or deny the testimony of an expert on the reliability of eyewitness identification, weighing a request to introduce such expert testimony "against other relevant factors, such as the centrality of the identification issue and the existence of corroborating evidence" (People v. Lee, 96 NY2d 157, 163 ). Because mistaken eyewitness identifications play a significant role in many wrongful convictions, and expert testimony on the subject of eyewitness recognition memory can educate a jury concerning the circumstances in which an eyewitness is more likely to make such mistakes, "courts are encouraged…in appropriate cases" to grant defendants' motions to admit expert testimony on this subject (People v. Drake, 7 NY3d 28, 31 , citing People v. Young, 7 NY3d 40 ).
In People v. LeGrand (8 NY3d 449 ), we set out standards governing the discretion of trial courts in regard to the admission of expert testimony on eyewitness identification. LeGrand established a two-stage inquiry for considering a motion to admit such testimony. The first stage is deciding whether the case "turns on the accuracy of eyewitness identifications and there is little or no corroborating evidence connecting the defendant to the crime" (LeGrand, 8 NY3d at 452). If the trial *11court finds itself with such a case, then it must proceed to the second stage, which involves the application of four factors. The court must decide whether the proposed "testimony is (1) relevant to the witness's identification of defendant, (2) based on principles that are generally accepted within the relevant scientific community, (3) proffered by a qualified expert and (4) on a topic beyond the ken of the average juror" (id.). If, on the other hand, sufficient evidence corroborates an eyewitness's identification of the defendant, then there is no obligation on the part of the trial court to proceed to the second stage of analysis, because testimony concerning eyewitness identifications is unnecessary (see e.g. Young, 7 NY3d at 45).
In LeGrand, one witness to a murder, who saw the perpetrator close up, identified the defendant as the killer in a photographic array and in a lineup, conducted seven years later. Two other witnesses, who identified the defendant at trial after they had failed to identify him in a photographic array, had seen his photograph in the array again in the district attorney's office the night before they testified. No physical evidence connected defendant to the killing. Defendant sought to introduce expert testimony on research concerning factors that may influence the perception and memory of eyewitnesses and the reliability of their identifications. The trial court, after a Frye hearing, precluded the testimony on the ground that the expert's conclusions were not generally accepted in the relevant *12scientific community.
We held that, with respect to much of the requested expert testimony on eyewitness identifications, the hearing court abused its discretion in failing to admit the testimony. In particular, we ruled that there was sufficient evidence to confirm that the principles underlying the expert's proposed testimony regarding the lack of correlation between confidence and accuracy of identification, confidence malleability, and the effect of postevent information on identification accuracy were generally accepted by social scientists and psychologists working in the field, and that testimony concerning these issues should not have been precluded (LeGrand, 8 NY3d at 458). By contrast, we ruled that there was insufficient evidence to confirm that the principles underlying the expert's testimony on weapon focus were generally accepted by the relevant scientific community, so that this testimony was properly excluded (id.).
We returned to the theme of expert testimony on eyewitness identification in People v. Abney and People v. Allen (13 NY3d 251 ). In Abney, a teenage girl was robbed of a necklace, on a subway station stairway, by a stranger wielding a large knife. The only evidence against defendant was the eyewitness identifications (in a photographic array and in a lineup) of the girl. Defendant requested permission to elicit expert testimony about the effects of event stress, exposure time, event violence and weapon focus, cross-racial *13identification, lineup instructions, and double blind lineups. In Allen, two masked men robbed a barbershop; one carried a knife. Two witnesses recognized the robber with the knife as an individual whom they knew from the neighborhood, and then identified defendant as that person in a photographic array. Defendant sought to elicit testimony from an expert concerning the effects of event stress, weapon focus, lack of correlation between confidence and accuracy, and unconscious transference. In both cases, expert testimony on eyewitness identifications was excluded.
In Abney, where "it was clear that there was no evidence other than [the victim's] identification to connect defendant to the crime, and she did not describe him as possessing any unusual or distinctive features or physical characteristics" (13 NY3d at 268), we held that the trial court abused its discretion in failing to allow expert testimony on the subject of witness confidence and in refusing to hold a Frye hearing with regard to the expert's proposed testimony on the effect of event stress, exposure time, event violence and weapon focus, and cross-racial identification.2
In Allen, on the other hand, the case against defendant did not depend exclusively on one eyewitness's identification, and we ruled that the trial court acted within its discretion.
*14"Allen is not a 'case [that] turns on the accuracy of eyewitness identifications [where] there is little or no corroborating evidence connecting the defendant to the crime' (LeGrand, 8 NY3d at 452). Critically, [a second eyewitness] independently identified defendant as the knife-wielding robber who searched him and stood nearby throughout the course of the robbery. And defendant was not a stranger to either [eyewitness]." (13 NY3d at 269.)
Whether a victim's or other eyewitness's identification of a defendant is sufficiently corroborated by other eyewitness identifications, so that the trial court need not proceed to the second stage of the LeGrand analysis, is dependent on the circumstances of the case. In LeGrand, no physical evidence linked the defendant to the crime, and the two doubtful eyewitness identifications did not sufficiently corroborate the identification of the eyewitness who saw the perpetrator close up. However, even when the evidence that a defendant was the perpetrator of a crime consists entirely of eyewitness identifications, the case is not necessarily one in which "there is little or no corroborating evidence" (8 NY3d at 452). Thus, in Allen, we held that the second eyewitness's identification of the defendant constituted evidence corroborating an eyewitness identification (13 NY3d at 269). We reached that conclusion because the corroborating identification possessed strong indicia of accuracy. In particular, the defendant in Allen was known to the second eyewitness, who recognized him during the robbery (id.).
*15In the present case, when Supreme Court denied Santiago's in limine motion in December 2003, the evidence disclosed by the People consisted of the victim's identifications of him in a photographic array and a lineup. At that time, as Supreme Court noted, the case turned on the accuracy of a single eyewitness identification and there was no corroborating evidence connecting the defendant to the crime. The issue therefore becomes whether the four factors enumerated in the second stage of the LeGrand analysis apply to the proposed testimony.
Supreme Court abused its discretion when it refused to allow testimony on studies showing that eyewitness confidence is a poor predictor of identification accuracy and on studies regarding confidence malleability. Undoubtedly, such testimony is relevant to this case in which the primary evidence against defendant—the only evidence at the time of Supreme Court's first ruling—was the victim's identification, and she expressed her recognition with subjective certainty. Moreover, the principles concerning confidence about which the expert was to testify are generally accepted within the relevant scientific community, and are beyond the ken of the average juror (Abney, 13 NY3d at 268; LeGrand, 8 NY3d at 458). For similar reasons, Supreme Court abused its discretion when it excluded expert testimony on the effects of postevent information on eyewitness *16memory (see LeGrand, 8 NY3d at 458).3
Given that the People did not dispute that the victim is a non-Hispanic Caucasian, the proposed testimony on inaccuracy of identifications of Hispanic people by non-Hispanic Caucasians appears relevant, and is beyond the ken of the average juror. Supreme Court should also have given more adequate consideration to whether the proposed testimony concerning exposure time, lineup fairness, the forgetting curve, and simultaneous versus sequential lineups was relevant to this case and beyond the ken of the average juror, and if necessary held a Frye hearing, to determine whether these factors are generally accepted as reliable within the relevant scientific community.4
By contrast, testimony concerning weapon focus, the effects of lineup instructions, wording of questions, and unconscious transference would have been irrelevant. The victim was not aware that her assailant had a weapon, and the record contains no evidence of improper lineup instructions, suggestive wording, or the presence of defendant's image in photographs the victim saw prior to identifying him in the photographic array she *17viewed. In these respects, exclusion was proper.
A separate question is whether Supreme Court abused its discretion when, after the defense had rested, the court denied defendant's renewed request to call an expert witness on eyewitness identification. By this time, the People had introduced evidence of Alarcon's and Rios's identifications of defendants, in addition to the victim's.
The People argue that, because of this additional evidence, this is not a case in which "there is little or no corroborating evidence connecting the defendant to the crime" (LeGrand, 8 NY3d at 452), and consequently it could not have been an abuse of discretion to exclude the requested expert testimony. We disagree.
Here, as in LeGrand, several factors call the corroborating identifications into question. Like the victim, Alarcon saw only part of the perpetrator's face. Alarcon identified defendant as the perpetrator with only 80percent confidence. It is also possible that Alarcon's December 2003 identification, using the photographs of the January 2003 lineup, was tainted by his memory of the photograph of defendant he had seen in the Spanish-language newspaper. Moreover, Rios's identification of defendant may have been influenced by his memory of the police artist's sketch of the assailant, calling into question the independence of this evidence from the victim's own *18identification. Taking into account all these circumstances, we do not consider the corroborating evidence sufficient to obviate the second stage of the LeGrand analysis.
Supreme Court, therefore, having erred in certain respects in denying defendant's pretrial motion to admit expert testimony regarding eyewitness testimony, erred in the same respects in denying defendant's end-of-trial motion to reopen the case and admit the expert testimony. Moreover, in its second ruling, Supreme Court should have given specific consideration to the proposed testimony concerning unconscious transference. That testimony would have been relevant, given that Alarcon saw a photograph of Santiago, and Rios saw a sketch of the perpetrator based on the victim's description, and familiarity with these images may have influenced these eyewitnesses' identifications.
Finally, Supreme Court's errors were not harmless. Trial error is only harmless when there is overwhelming proof of the defendant's guilt and no significant probability that the jury would have acquitted the defendant were it not for the error (People v. Crimmins, 36 NY2d 230, 242 ). Here, the proof of defendant's guilt was not overwhelming; therefore, the errors cannot be regarded as harmless. We need not decide the probability that the verdict would have been different if the expert testimony had not been excluded.
Accordingly, the order of the Appellate Division should be reversed and a new trial ordered.
1. We decided otherwise, with respect to postevent information, in People v. LeGrand, 8 NY3d 449, 458 ).
2. However, we ruled that expert testimony on lineup instructions and double-blind lineups would have been irrelevant to the victim's identification of defendant.
3. While we had not yet decided LeGrand at the time Supreme Court reached its decision, the trial court should have held a Frye hearing to investigate the scientific acceptability of the principles related to eyewitness confidence and postevent information.
4. We note that the fact that the victim assisted in the creation of an artist's sketch of her attacker does not render the expert testimony irrelevant.
*19Order reversed and a new trial ordered. Opinion by Judge Pigott.Chief Judge Lippman and Judges Ciparick, Graffeo, Read, Smith and Jones concur.