I agree that reversal is not required under the rule of People v LeGrand (8 NY3d 449 [2007]), but disagree with the analyses of Justices Buckley and Moskowitz.
*174The rule of LeGrand, of course, is that
“where [a] case turns on the accuracy of eyewitness identifications and there is little or no corroborating evidence connecting the defendant to the crime, it is an abuse of discretion for a trial court to exclude expert testimony on the reliability of eyewitness identifications if that 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. at 452).
This case certainly turns on the accuracy of eyewitness identification testimony, and the expert’s testimony satisfies, at least with respect to the subject of witness confidence and accuracy, each of the above four factors.1 Is this case, however, one in which there is “little or no corroborating evidence connecting the defendant to the crime”? If it is, can the erroneous exclusion of the expert’s testimony be considered harmless error? In my view, we need not answer the first question; but assuming the answer is yes, the answer to the second question is yes.
In the context of this case, the first question reduces to the following: if the evidence at trial is that more than one eyewitness identified the defendant, can the identifications cross-corroborate each other so that the case is not one in which there is but “little or no corroborating evidence connecting the defendant to the crime?” The recent decision of the Court of Appeals in People v Abney (13 NY3d 251 [2009]), suggests that the answer is yes.2 In People v Allen, the other case decided with People v Abney, two eyewitnesses, Almonte and Bierd, identified the defendant at lineups held four months after the crime and at trial (id. at 262-263). The Court ruled that the case “did not depend exclusively on Bierd’s eyewitness testimony—i.e., [it] is *175not a case [that] turns on the accuracy of eyewitness identifications [where] there is little or no corroborating evidence connecting the defendant to the crime” (id. at 269 [internal quotation marks omitted; brackets in original]). A unanimous Court explained: “Critically, Almonte 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 Bierd or Almonte” (id.).
As is evident, two considerations informed the Court’s conclusion: the “independent[ ]” identification by Almonte, and that defendant was known to both identifying witnesses. Although the second consideration might be thought to provide more powerful support (If a defendant is known to the identifying witness or witnesses, can such a case be one that turns on the accuracy rather than the honesty of eyewitness identification testimony?), the Court stated that the first consideration was “[c]ritical[ ]” to its conclusion. Regardless of the relative importance of the two considerations, at the very least the opinion suggests that multiple corporeal identifications that are “independent[ ]” can cross-corroborate each other so that the case is not one in which there is but “little or no corroborating evidence connecting the defendant to the crime.”3
The dissent is unpersuasive in concluding otherwise, although I need not decide whether the dissent is wrong. In the first place, the analysis is not “quite simple,” because it is not at all clear how much and what forms of corroboration are sufficient to provide more than “little . . . corroborating evidence connecting the defendant to the crime” (LeGrand, 8 NY3d at 452). Of course, the dissent is correct that “the problem of misidentification can exist whether there is one eyewitness or several.” *176But the question is whether all cases of multiple identifications, regardless of their particular facts, must be treated the same for the purpose of the LeGrand rule. Contrary to the dissent, the opinion in LeGrand does not purport to address the question. True, the opinion notes that three witnesses identified the defendant during the first trial (8 NY3d at 453). But the Court immediately went on to note that “two of the witnesses had seen defendant’s photo array in the district attorney’s office the night before they were to testify” (id.). Thus, the Court may not have regarded these corporeal identifications as “independen[ ].” It seems unlikely, moreover, that the Court intended in LeGrand to decide without discussion the important question of whether multiple corporeal identifications can provide the corroboration connecting the defendant to the crime that can justify a trial court’s decision to exclude the testimony of a defense expert. Finally, the dissent fails to deal with or even mention the statement in Abney (Allen), quoted above, that a “[c]ritical[ ]” factor supporting the Court’s conclusion that the exclusion of the expert testimony did not violate the LeGrand rule was that “Almonte independently identified defendant as the knife-wielding robber.” Rather than regard the statement as gratuitous for some reason, I would take the Court at its word.
Justice Buckley is unpersuasive to the extent his position is that under LeGrand the testimony of a defense expert properly can be excluded “where there [is] reliable witness identification testimony” (emphasis added). Nothing in LeGrand or Abney (Allen) (nor, for that matter, in People v Lee [96 NY2d 157 (2001)] or People v Young [7 NY3d 40 (2006)]) supports the proposition that trial courts properly can exclude expert testimony offered by the defense when the identification testimony from the prosecution can be regarded as “reliable.” Nor do the decisions of this Court cited by Justice Buckley so hold. In People v Smith (57 AD3d 356 [2008], lv denied 12 NY3d 821 [2009]), the panel noted the “highly reliable multiple eyewitness identifications” (id. at 357), but did not rest its holding on that ground. Rather, the holding was that “the exclusion of expert testimony on the reliability of eyewitness identification” did not require reversal “[s]ince there was sufficient corroboration of defendant’s guilt, including consciousness-of-guilt evidence and partially incriminating statements to the police” (id.). Similarly, People v Chisolm (57 AD3d 223 [2008], lv denied 12 NY3d 782 [2009]) upheld the exclusion of expert testimony *177on eyewitness identification because “[t]here was significant corroborating evidence of defendant’s guilt” (id. at 223-224 [citation omitted]). In People v Austin (46 AD3d 195 [2007], lv denied 9 NY3d 1031 [2008]), the panel relied on preservation grounds when it ruled that the defendant had “failed to establish at any point [at trial] the relevance of the proffered testimony to the particular facts of this case” (id. at 199). Accordingly, the subsequent discussion (id. at; 200-201), contrasting the factual circumstances of the identification with those in LeGrand, appears to be dicta.
We should not try to read the tea leaves in LeGrand and Abney and determine whether multiple identifications can cross-corroborate each other so as to provide more than a “little . . . corroborating evidence connecting the defendant to the crime.” Regardless of how the Court of Appeals may tule on that question, I think it clear the erroneous exclusion of expert testimony under LeGrand is subject to harmless error analysis and that any error in this case was harmless.4
*178At first blush it may seem odd to think that harmless error analysis has any role to play when a trial court abuses its discretion by excluding expert testimony offered by the defense in a case where there is “little or no corroborating evidence connecting the defendant to the crime.” If there is “little or no corroborating evidence,” how can the error be harmless? But cases can readily be conceived (including ones like Abney [Allen] in which the defendant is known by the identifying witness or witnesses and those noted earlier in which there are corporeal identifications by multiple witnesses each of whom had extended opportunities to view the perpetrator) in which it would be incomprehensible not to permit harmless error analysis to play a role. Moreover, Abney all but states that harmless error analysis does apply (13 NY3d at 268 [“we do not consider the trial judge’s error in Abney to have been harmless”]). In addition, the Legislature has commanded appellate courts in peremptory terms to determine appeals “without regard to technical errors or defects which do not affect the substantial rights of the parties” (CPL 470.05 [1]). On this score, I also note that this certainly is not a case in which the Federal Constitution precludes harmless error analysis (see generally United States v Gonzalez-Lopez, 548 US 140, 148-151 [2006] [discussing “structural defects,” a class of constitutional error that, as opposed to “trial error,” is not susceptible to harmless error analysis]).
In the first place, the victim provided the jury with a compelling basis for concluding that the People had proven defendant’s identity as the attacker beyond a reasonable doubt. This is not a case in which the identifying witness had but a fleeting opportunity to view an attacker under stressful circumstances. To the contrary, the victim first saw the man who would attack her when he was at the opposite end of the platform. As the People argue, despite the distance and the want of any particular reason to pay attention to him, the victim observed the man well enough to recognize him when, shortly thereafter, she saw him again from a closer distance. On that second occasion, she took note of the man’s layered clothing and unusual behavior of making eye contact and then stepping behind pillars. She then had an extended opportunity to observe her attacker in good lighting before he assaulted her. That is, after he approached her, she looked at him from about an arm’s length distance for several seconds while waiting for him to speak. When he asked her if she was “working,” the two were face-to-face; the ensu*179ing conversation lasted for about 30 seconds, until she was attacked. Although her attacker had a garment covering his mouth and chin, the victim could see the man’s face from his moustache to above his eyebrows. Moreover, as the People correctly maintain, in describing her attacker to the police, the victim consistently provided details that closely matched defendant’s ethnicity, complexion, age and facial hair. Thereafter, she viewed thousands of photographs, and never identified anyone until she saw, only about two weeks after the attack, a recent photograph of defendant in an array of photographs of Hispanic men with moustaches. When she saw the photograph of defendant, her “heart stopped,” the same reaction she had just two days later when she identified defendant at a lineup.
As is clear from Justice Buckley’s writing, there was much more evidence of defendant’s identity as the attacker. With respect to Alarcon, suffice it to say that when he first saw the attacker, as the man walked past him toward the victim, the attacker had not yet covered his mouth and goatee. Alarcon saw him again shortly thereafter, when he heard the victim’s screams and saw the attack. Of course, Alarcon testified that he was only 80% sure of his identifications of defendant (the undeclared identification at the lineup and the in-court identification). But his testimony corroborated the victim’s identification nonetheless, and that same testimony gave the jury a sound reason for concluding that Alarcon was an honest and careful witness. Finally, Rios also got good looks at the attacker’s face, before and after the attack, and also gave a description of the attacker that fit defendant with respect to ethnicity, complexion, age and facial hair. Like the victim, Rios was certain of his identification of defendant as the attacker.
Regardless of whether the constitutional standard of harmless error might apply in another case, the nonconstitutional standard applies here because defendant never alerted the trial court to any claim of constitutional error with respect to the expert’s testimony (People v Kello, 96 NY2d 740, 743-744 [2001]).
For all these reasons, and for those cited by Justice Buckley in support of his position that “it was reasonable for the trial court to conclude that the identification by multiple, corroborative witnesses ‘was quite unlikely to be mistaken,’ ” I conclude both that the evidence of guilt was overwhelming and that there is no significant probability that the jury would have acquitted if the expert’s testimony had not been excluded (Kello, 96 NY2d *180at 744). Moreover, in its final charge to the jury, the court admonished the jurors 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.” That authoritative instruction provided defendant with at least some of the benefit of the expert’s testimony and further attenuates the possibility that the verdict might have been different if the expert’s testimony had not been excluded.
The dissent apparently agrees that harmless error analysis is applicable and that the appropriate standard in this case is the nonconstitutional standard. Oddly, however, the dissent nonetheless relies, inter alia, on the fact that no property of the victim was found with defendant when he was arrested5 and that none of the witnesses had “ever seen him before.” Of course, however, if property of the victim had been found with defendant or if one of the identifying witnesses knew defendant independently of the crime, there would be ample corroborating evidence connecting defendant to the crime and thus no LeGrand error.
The dissent is wrong in asserting that “[n]one of the witnesses had an extended opportunity to view the assailant.” The victim had such an opportunity—under good lighting conditions—and it included a 30-second period of conversation before the attack began, with the assailant standing right in front of the victim. True, the other two identifying witnesses did not observe the attacker for as extended a period. But they, too, observed the attacker under good lighting conditions and saw him both before (i.e., under nonstressful circumstances) and after the attack. Although the opinion in Abney states only that the entire encounter between the victim and the assailant was “fleeting” (13 NY3d at 257), the dissent nonetheless states that the 30-second, pre-attack conversation the victim in this case had with her attacker “appears not to be much longer.” In any event regardless of our disagreement about whether the preattack encounter provided the victim with an “extended” opportunity to view her attacker, the encounter certainly was not one in which the victim caught only a fleeting glimpse of him. The ability of the victim to provide so many descriptive details of her attacker also stands as persuasive proof of the reliability of her identification (see People v Huertas, 75 NY2d 487, 492 *181[1990] [description given by identifying witness “was probative of her ability to observe and remember her assailant, and thus relevant to the accuracy of the identification she made”]).
The dissent does not come to grips with these facts. Nor does it come to grips with the accuracy of the descriptions given by the victim and Rios, the fact that the lineup occurred just two weeks after the attack or the caution about witness confidence given by the court in its final charge to the jury. Indeed, even though it does not mention that the victim recognized defendant as her attacker at a lineup just two weeks after the attack, the dissent relies on the ostensible fact that Alarcon did not “identif[y] defendant as the assailant until nearly a year after the crime occurred.” But Alarcon did see and recognize defendant at the same lineup; all that the dissent fairly can say is that he did not declare his identification until considerably later. Above all, however, the dissent fails to come to grips with the fact that the identifications of each of the three identifying witnesses cross-corroborated each other.6
. With respect to the other proposed subject of the expert’s testimony that is disputed on appeal—the so-called “forgetting curve”—I agree with the People’s position that (1) it is common knowledge that memory generally becomes less reliable over time, and (2) while the specific principle that memory loss is greatest immediately after an event may be beyond the ken of the average juror, that principle was not supported by any specific studies identified in defendant’s proffer (see People v Carrieri, 49 AD3d 660 [2d Dept 2008], lv denied 11 NY3d 786 [2008])
. A recent article explores significant issues arising from LeGrand and Abney (see Paul Shechtman, In the Area of Eyewitness Identification Expert Testimony, LeGrand Should Be Revisited, 8 NY Crim L Newsl [No. 2] 8 [NY St Bar Assn. Spring 2010]).
. Circumstances can be conceived in which it would be particularly difficult to defend the conclusion that multiple corporeal identifications never can cross-corroborate each other for purposes of the LeGrand rule. Suppose, for example, three eyewitnesses, each of whom had extended opportunities to observe the perpetrator under nonstressful circumstances, separately gave accurate and detailed descriptions and identified him at separate lineups shortly after the crime and at trial. Of course, the proposed expert testimony might undercut the force of the identification testimony of each identifying witness. According to the People, however, even the defense expert who testified at the Frye hearing in LeGrand conceded that the risk of a false identification can be “substantially reduced if two or more witnesses are available” (People’s Brief in People v LeGrand at 102). On the other hand, the conclusion that multiple corporeal identifications never can cross-corroborate each other would be easier to defend if harmless error analysis is applicable to a violation of the LeGrand rule. After all, these hypothetical circumstances also would support a harmless error argument.
. Deciding this case on harmless error grounds obviates the need to wrestle with constitutional questions raised by Holmes v South Carolina (547 US 319 [2006]). There, the question was “whether a criminal defendant’s federal constitutional rights are violated by an evidence rule under which the defendant may not introduce proof of third-party guilt if the prosecution has introduced forensic evidence that, if believed, strongly supports a guilty verdict” (id. at 321). The Court unanimously ruled that the evidence rule was unconstitutional, explaining that “[t]he point is that, by evaluating the strength of only one party’s evidence, no logical conclusion can be reached regarding the strength of contrary evidence offered by the other side to rebut or cast doubt” (id. at 331). LeGrand, of course, does not hold that the testimony of a defense expert is inadmissable where the prosecution’s evidence can be regarded as strong (i.e., supported by more than a “little . . . corroborating evidence connecting the defendant to the crime”). But LeGrand does hold that if the identification testimony adduced by the prosecution is sufficiently corroborated, a trial court does not abuse its discretion when it excludes the testimony of the defense expert. Whether that holding is consistent with Holmes v South Carolina is unclear. Nor is it clear whether a rule that permits the exclusion of the testimony of a defense expert when the prosecution has presented proof of multiple and independent identifications (or when the identification testimony presented by the prosecution is considered sufficiently reliable) would pass muster under Holmes v South Carolina. One significant question, however, comes into focus when it is considered that under LeGrand criminal defendants are entitled to an opportunity to persuade the jury of a reasonable doubt through the testimony of an expert when there is “little or no corroborating evidence” connecting them to the crime. Can a defendant be deprived of that opportunity whenever the prosecution’s evidence is marginally stronger, when there is some incremental evidence so that there is more than a “little” corroborating evidence?
. As the attacker did not steal anything from the victim, it hardly is surprising that defendant was not in possession of any of the victim’s property when he was arrested.
. I repeatedly make clear that I do not and need not decide the question of whether multiple corporeal identifications can cross-corroborate each other so that the exclusion of a defense expert’s testimony is not error under LeGrand. Moreover, I also make clear that I believe there are good reasons not to decide the question. Nonetheless, the dissent apparently believes that I have decided that question.