Judicial Scrutiny of Eyewitness Evidence: Lessons From the 2024 Court of Appeals Decisions
6.30.2025
As a matter of course, court decisions on major issues evolve over time, in piecemeal fashion. The reasons are obvious: cases begin at a lower court and are filed at different times under different circumstances, while decisions are limited to the facts at hand, often leaving questions about broader applicability to be decided by higher courts. This has certainly been the case over the past several decades, during which the New York Court of Appeals has developed a cautious and evolving body of law around the use of eyewitness identification evidence – generally through individual rulings addressing specific procedural or evidentiary issues. During its 2024 term, the court returned to this topic with unusual focus, issuing four decisions – Vaughan, Williams, Mosley, and Watkins – that clarify key principles governing expert testimony, in-court identifications, lay video identifications, effective assistance of counsel, and cross-race jury instructions. While none of the rulings marks a seismic shift, together they offer a more cohesive framework for evaluating the reliability of eyewitness evidence – one that reflects lessons from wrongful conviction data, developments in cognitive science, and longstanding constitutional protections.
Eyewitness misidentification has long been recognized as the primary contributing factor to wrongful convictions in the United States, appearing in approximately 70% of DNA exonerations. Even before this reality was confirmed through empirical study and DNA testing, the New York Court of Appeals acknowledged the inherent fallibility of eyewitness memory. Over the decades, the court has crafted a body of case law aimed at reducing the risks associated with eyewitness identifications, culminating in a series of important rulings during the 2024 term.
Expert Eyewitness Testimony Admissible Even Where Corroborating Evidence Exists
The court’s 2007 decision in People v. LeGrand[1] represented a major turning point in the Court of Appeals’ eyewitness jurisprudence in that it permitted expert testimony about eyewitness memory. In LeGrand, the court held that where the case turns on a contested identification and there is little or no corroborating evidence linking the defendant to the crime, trial courts must allow expert testimony on factors affecting the reliability of eyewitness identification if it would assist the jury.
LeGrand was understood to create a two-step test for the admission of expert eyewitness testimony and was often interpreted to exclude experts where corroborating evidence appeared to exist. This interpretation raised practical and constitutional concerns.
First, the Supreme Court has made clear that a defendant’s right to present a complete defense – rooted in the due process clause and the Sixth Amendment – prohibits trial courts from excluding relevant defense evidence simply because the government’s case appears compelling.[2] The right to present a defense includes the ability to challenge the credibility and reliability of prosecution evidence. Allowing expert testimony only where the government’s case appears weak – i.e., where there is a little to no corroborating evidence – would appear to violate this mandate.
Second, in many wrongful conviction cases, the very evidence courts later deem corroborative – such as matching clothing, proximity to the scene, or prior acquaintance – is what caused the innocent person to become a suspect in the first place. Once a name or image is introduced to an eyewitness, particularly through suggestive means, eyewitnesses may rely more on these contextual cues than on genuine facial recognition from memory.[3] This phenomenon, known as unconscious transference or pre- or post-procedure feedback, can lead witnesses to misattribute familiarity or other associative details to the person they are asked to identify. Thus, the apparent strength of corroboratioIndependent n may not only fail to protect against misidentification but may actually contribute to it.
Third, at the time a court is considering a defense motion for an expert, the prosecution’s evidence has yet to be meaningfully tested – so what appears to be strong corroboration pre-trial may wind up being weak or not credible evidence. Indeed, the known wrongful conviction data (from DNA exonerations) makes this clear. Mistaken eyewitness identification played a contributing role in nearly 70% of the more than 375 DNA-based exonerations in the United States. In many of these cases, courts and juries were persuaded by what appeared to be strong corroboration – even after trial – multiple eyewitnesses misidentifying the same person, matching clothing, or a similarity between the innocent defendant and the suspect. Yet these very elements often proved to be unreliable or misleading, as later DNA testing demonstrated.
In People v. McCullough,[4] the court “clarified” that LeGrand should not be interpreted as mandating a rigid two-part test focused solely on the strength of corroborating evidence. Instead, the court explained, LeGrand enumerated factors for trial courts to consider when determining whether such expert testimony would assist the jury.
In its most recent decision on the admissibility of expert testimony, People v. Vaughan,[5] the New York Court of Appeals further “clarified” LeGrand, holding that the existence of corroborating evidence does not preclude the admission of eyewitness expert testimony. The court explained that corroborative evidence, while relevant, is not dispositive of whether expert testimony on eyewitness reliability should be admitted. Rather, trial courts must conduct a fact-specific inquiry to determine whether the proposed expert testimony would assist the jury in evaluating the identification. The court emphasized that trial courts must avoid short-circuiting the analysis by relying solely on the perceived strength of the prosecution’s evidence.
Specifically, courts must engage in a case-specific assessment of whether expert testimony would assist the jury in evaluating the identification’s reliability – even when circumstantial or contextual corroboration exists. Perhaps most significantly, Vaughan reaffirms that trial courts retain discretion over the admissibility and scope of expert testimony but cautions that this discretion must be exercised thoughtfully. Judges may not substitute their own assessment of the strength of the prosecution’s case for a reasoned analysis of whether the jury would benefit from expert insight into eyewitness memory. In this way, Vaughan guards against the rigid or overly limiting application of LeGrand, preserving the trial court’s gatekeeping role while protecting the defendant’s right to a fair trial.
Although Vaughn did not directly address the constitutional issues, its rejection of a categorical rule barring expert testimony in the face of corroboration aligns with the principles set forth in Crane v. Kentucky. Judicial gatekeeping must not override the jury’s role in assessing reliability and guilt.
Vaughan affirms trial court discretion while clarifying that corroborative evidence alone cannot justify the exclusion of expert testimony on eyewitness reliability. In doing so, it harmonizes New York law with evolving scientific understanding and constitutional principles. The decision strengthens protections against wrongful convictions rooted in mistaken identification and ensures that juries are better equipped to evaluate one of the most fallible forms of evidence in the criminal justice system.
Separate Independent Source Hearings and Out-of-Court Identification
In People v. Williams,[6] the Court of Appeals addressed a recurring issue in undercover narcotics operations: whether an in-court identification by an officer may be admitted when an earlier out-of-court identification has been suppressed as the fruit of an unlawful arrest. The decision reinforces the procedural requirement that courts conduct a formal independent source hearing before admitting any such in-court identification.
The case arose from a buy-and-bust operation in which an undercover officer observed a drug sale and described the alleged seller – whom he had not previously encountered – as a Black male wearing black pants, a white sweater, and a black hat. Another officer subsequently arrested the defendant based on this description. At a probable cause hearing, the trial court found the arrest unlawful and suppressed both the out-of-court identification and physical evidence. Nevertheless, the court allowed the undercover officer to identify the defendant in court, concluding sua sponte that there was an independent source for the identification based solely on the officer’s probable cause hearing testimony.
The Court of Appeals reversed. It held that once an identification has been suppressed as a product of unlawful police conduct, the trial court must hold a separate, formal, independent source hearing – at which the prosecution bears the burden of proving by clear and convincing evidence that the in-court identification derives from observations independent of the tainted identification. The court emphasized that such a hearing is not optional, nor may the trial court rely solely on prior hearing testimony to make this determination.
Importantly, the court rejected the notion that police officers are specially equipped to evaluate the source of their own identifications. It cautioned that “the nature of a police officer’s duties, special training, and experience on their own simply have no relevance” to whether an identification is the product of illegal conduct. This language underscores the principle that police credibility or expertise does not exempt law enforcement witnesses from scrutiny under constitutional safeguards.
The court outlined a set of factors that must be considered at an independent source hearing, including:
- The witness’s opportunity to observe the defendant at the time of the crime.
- The accuracy and detail of the initial description.
- The degree of certainty expressed at the time of the illegal identification.
- The level of prior familiarity (if any).
- The time elapsed between the observation and the identification.
- The nature and circumstances of the suppressed identification and the risk of taint.
Williams thus reinforces procedural rigor in identification cases, particularly where police-initiated identifications may be tainted by unconstitutional arrests. The decision reaffirms that the reliability of such identifications must be scrutinized through a structured evidentiary process – not assumed based on the witness’s professional status or the perceived strength of the prosecution’s case.
A New Test for Lay Non-Eyewitness Identifications
In People v. Mosley,[7] the Court of Appeals addressed the increasingly common practice of using law enforcement officers – and other lay witnesses – to identify a defendant as the person depicted in surveillance footage, despite not having witnessed the crime. The court reversed the conviction, finding that the trial court erred in permitting a police officer to offer identification testimony without first establishing that the officer had sufficient familiarity with the defendant to render that opinion helpful to the jury.
The court emphasized that non-eyewitness identification testimony – especially from law enforcement – poses unique risks. First, such testimony may usurp the fact-finding function of the jury. Second, jurors may give undue weight to such testimony, particularly when offered by authoritative figures, even if it is based on brief or limited exposure to the defendant. With respect to law enforcement witnesses, the court specifically noted that “[t]rial testimony by a law enforcement officer may pose additional concerns by drawing attention to a defendant’s prior interactions with the police, and efforts to mitigate any prejudice by omitting reference to a law enforcement officer’s occupation or providing a cryptic basis for familiarity may constrain the opportunity for vigorous cross-examination.”[8]
In Mosley, the court found that there had been no showing that the officer witness – who was not an eyewitness to the crime and was merely called to identify the defendant as the person in the surveillance video showing the crime – was familiar enough with the defendant’s appearance to offer a reliable opinion, nor was there any indication that the jury needed assistance interpreting the surveillance video. The defendant had not altered his appearance in a way that would have impeded the jury’s independent assessment, and the video quality did not warrant lay interpretation
In distinguishing Mosley from its prior decisions in People v. Russell[9] and People v. Sanchez,[10] the court underscored two critical limitations. In those earlier cases, lay identification testimony was permitted because the witness had prior personal knowledge of the defendant and the defendant had used a disguise during the crime, thus impairing the jury’s ability to assess the video unaided. Neither condition was met in Mosley.
The court adopted a two-part test governing the admissibility of non-eyewitness identification testimony:
- Familiarity: The proffered witness must have had “sufficient contact” with the defendant to achieve a level of familiarity that renders their identification helpful. If the witness lacks such familiarity, the jury should make its own conclusion without the witness’s assistance. Factors to consider include: the witness’s general level of familiarity with the defendant’s appearance; the span of time and variety of circumstances of the witness’s observations of the defendant; the witness’s familiarity with the defendant’s appearance at the time the video was taken; and any familiarity with the defendant’s customary manner of dress or clothing on the day the video was recorded. Finally, courts should consider whether the witness references a specific trait the defendant has and whether that trait is apparent in the video footage.
- Necessity: The trial court must determine whether the jury requires assistance to identify the defendant in the video – such as where the defendant has altered their appearance or where the footage is of low quality or obstructed.
Both elements must be established outside the jury’s presence during a voir dire or pretrial hearing, with the proponent bearing the burden of demonstrating that the testimony would be both helpful and necessary. The court urged both parties to create a thorough record to aid the court in its determination and to allow for meaningful appellate review.
Finally, as a best practice, the court urged trial courts to provide cautionary jury instructions, both at the time of the testimony and during the final charge, explaining that lay non-eyewitness identification testimony is “mere opinion testimony that they may choose to accept or reject” and reminding jurors that it is their opinion that controls.
Counsel May Be Ineffective for Failing To Request a Mandatory Cross-Race Jury Instruction
In People v. Watkins,[11] the New York Court of Appeals addressed whether defense counsel’s failure to request a cross-race jury instruction constituted ineffective assistance of counsel. While the court ultimately rejected the defendant’s claim, the decision reaffirms the importance of such instructions and highlights their growing role in safeguarding against misidentification.
The defendant, a Black man, was identified by a non-Black eyewitness in a single-witness case. Defense counsel did not request a cross-race identification charge, and none was given. On appeal, the defendant argued that this failure constituted constitutionally deficient performance under People v. Baldi[12] and Strickland v. Washington.[13]
The court acknowledged that cross-race identifications raise unique risks of misidentification, as extensively documented in psychological research. However, it concluded that at the time of trial – before the court’s landmark ruling in People v. Boone[14] – there was no clearly established duty to request such a charge. In Boone, the court held that upon request, trial courts must provide a jury instruction explaining the potential unreliability of cross-race identifications. Because Watkins predated Boone, the court found that counsel’s performance, while perhaps suboptimal, was not legally deficient under the standards prevailing at the time.
Notably, the court left open the possibility that a failure to request a cross-race instruction after Boone could support an ineffective assistance claim. It stressed the significance of the Boone rule and signaled that counsel’s awareness of cross-race identification risks is now part of the baseline competence expected of criminal defense attorneys.
Watkins serves as a cautionary reminder: in any case involving a cross-racial identification, defense counsel must request an appropriate instruction or risk a post-conviction claim. The decision reinforces the broader principle that effective assistance of counsel includes attention to well-established factors affecting eyewitness reliability – particularly when jury instructions are available to mitigate those risks.
Conclusion
The 2024 term reflects the New York Court of Appeals’ continued and evolving concern with the risks eyewitness identifications pose to the integrity of criminal trials. In Vaughan, Williams, Mosley, and Watkins, the court addressed the reliability of identifications from multiple angles – clarifying when expert testimony is required, limiting the admissibility of law enforcement identifications absent sufficient procedural safeguards, and reinforcing the constitutional necessity of tailored jury instructions.
Taken together, these decisions underscore the court’s recognition that traditional indicia of reliability – such as law enforcement expertise, multiple witnesses, or corroborative detail – may not suffice when viewed against decades of wrongful conviction data and social science. They also emphasize the importance of evidentiary rigor: that identifications must be tested, that expert guidance should be provided when needed, and that juries must be properly instructed on the cognitive limitations that affect eyewitness accuracy.
For practitioners, these rulings reinforce several practical imperatives: move early to request expert testimony where identification is contested; challenge lay and law enforcement identifications lacking foundational reliability; demand cross-race jury instructions in appropriate cases; and preserve a full record for appellate review. The court’s message is clear – protecting the fairness of criminal trials requires close scrutiny of identification evidence at every stage.
Karen A. Newirth is a founding partner of Newirth Linehan, where she represents wrongfully convicted individuals in post-conviction and civil rights litigation. She previously led systemic litigation and reform efforts at the Innocence Project. She is co-author of “Eyewitness Testimony: Civil and Criminal” and frequently writes and lectures on criminal legal reform.
Endnotes
[1] 8 N.Y.3d 449 (2007).
[2] Crane v. Kentucky, 476 U.S. 683, 690 (1986).
[3] See, e.g., People v. Abney, 13 N.Y.3d 251 (2014).
[4] 27 N.Y.3d 1158 (2016).
[5] 2024 N.Y. Slip Op. 01345 (Feb. 22, 2024).
[6] People v. Williams, 41 N.Y.3d 551 (2024).
[7] 41 N.Y.3d 640 (2024).
[8] Id. at 647 (cleaned up).
[9] 79 N.Y.2d 1024 (1992).
[10] 21 N.Y.3d 216 (2013).
[11] 42 N.Y.3d 635 (2024).
[12] 54 N.Y.2d 137 (1981).
[13] 466 U.S. 668 (1984).
[14] 30 N.Y.3d 521 (2017).
