GLICKMAN, Associate Judge:

Elliot Heath appeals his convictions for armed first-degree pre-meditated murder, armed assault with intent to kill, and associated weapons offenses. He primarily contends that the trial court unconstitutionally deprived him of a meaningful opportunity to present a complete defense when it erroneously excluded expert testimony he sought to present on the subject of eyewitness identification. The government concedes, and we agree, that the court erred in failing to conduct the inquiry our cases have held necessary to support its exclusionary ruling. Even so, considering appellant's proffer, we see no reasonable probability that the excluded expert testimony would have led the jury to have a reasonable doubt of appellant's guilt that did not otherwise exist. We therefore are persuaded that appellant's constitutional right to present a defense was not violated by the exclusion of his expert's testimony, and we conclude that the trial court's error was harmless.

I.

The charges against appellant arose from the murder of Patrick Carter on the morning of November 17, 2005, as he sat with his girlfriend, Felicia Edwards, in his parked car outside his mother's house in the 1800 block of Corcoran Street, N.E. In the moments before the shooting took place, Ms. Edwards was entering information into the calendar of her mobile phone. She happened to look up and noticed two men standing near the bus stop at the end of the street, roughly 250 to 350 feet away. When she looked up again, she saw them walking down Corcoran Street in her direction. There appeared to be nothing amiss. Ms. Edwards did not recognize either individual. She later described one of them to police as a tall, dark-complected black man with hair in dreadlocks, and the other as shorter, lighter-skinned, and bald.

Unexpectedly, the two men stopped next to the car in which Mr. Carter and Ms. Edwards were sitting. Then, without warning, the men began shooting into the vehicle. Mr. Carter was hit seven times and mortally wounded. As Ms. Edwards reached across to shield Carter and try to start the car, she too was shot, though not fatally. After the shooting, the assailants fled. Their motive for the attack is unknown.

Appellant was implicated in the shooting eight months later, when two apparently unrelated events led to his identification and arrest. The first event occurred on July 20, 2006, when a woman named Courtnee Ervin, arrested for violating bail on pending drug and prostitution charges, told detectives she had witnessed the shooting. Ms. Ervin explained that she was present in the 1800 block of Corcoran Street, getting high on crack cocaine, when the shooting erupted. She looked up and, before fleeing, got a good look at the two shooters. Ms. Ervin was positive that one of them was "L," a person she had known for some years from her frequent visits to Corcoran Street. When the detectives then showed Ms. Ervin a photograph of appellant, she confirmed that he was the "L" she knew.

The second event leading to appellant's arrest occurred about a week after Ms. Ervin spoke to the police, when a tall, dark-skinned man with dreadlocks walked into the shoe store where Ms. Edwards worked as a salesperson. The man was a stranger to Ms. Edwards, but upon seeing him, she became very nervous and felt physically ill. She was so distressed that she ran into a back room and called both her parents and a police detective working on the investigation of the Corcoran Street shooting to report her sighting of the man. Two days later, on July 29, 2006, the detective showed Ms. Edwards an array of nine photographs of African-American men of approximately the same age and complexion, all with their hair in dreadlocks. Included in the array was the photo of appellant that Ms. Ervin had said depicted the shooter she knew as "L." Ms. Edwards selected that same photo, saying it portrayed the man who had entered her store. At that time, she did not identify appellant as one of the men who had attacked her and Mr. Carter. However, when Ms. Edwards appeared before the grand jury a few months later, in November 2006, she had become certain that appellant was one of the assailants. In her subsequent testimony at trial, Ms. Edwards referred to him as "L," the detectives having told her his nickname.

The government's case against appellant rested heavily on the mutually reinforcing identification testimony of Ms. Edwards and Ms. Ervin, neither of whom knew or had spoken with the other. A third witness corroborated their testimony. Danielle Carter, the decedent's sister, testified at trial that she was inside her mother's house on Corcoran Street when she heard what sounded to her like firecrackers. She looked out the second-story window and saw a man running away from her brother's car. Not having seen the fleeing man's face, Ms. Carter said she could not identify him definitively. Nonetheless, she thought he was appellant, a person whom she had seen before on Corcoran Street "almost every day," because he had the same height, build, gait, and "shoulder-length dreads" as appellant had, and because he was wearing the same kind of dark winter coat she had seen appellant wear.

In his defense, appellant denied any involvement in the shooting and claimed he had been misidentified — either mistakenly, in the cases of Ms. Edwards and Ms. Carter, or intentionally, in the case of Ms. Ervin. Appellant cross-examined Ms. Edwards extensively, focusing on her inability to see either assailant's face during the attack because her view was blocked by the roof of the car, and on her three-month delay in identifying him as one of the shooters after having observed him in her store. Ms. Edwards admitted telling the police that she "really didn't get to see" the shooter with the dreadlocks and explaining to the grand jury that she saw only "bits and pieces" of that individual. She acknowledged, moreover, that her nervous reaction when she encountered appellant in her store was not unusual, as she became upset and anxious in the months following the shooting when she saw other dark-complected men with dreadlocks. (She said her reaction to appellant was more intense, however.) Appellant probed whether Ms. Edwards's belated and sudden identification of him a year after the shooting was influenced by the recurrent nightmares she reported having experienced in the interim, in which, she said, she had relived the traumatic event and reconstructed its details.

In cross-examining Ms. Ervin, appellant impeached her with her prior convictions and with her inconsistent statements to the police and the grand jury regarding her location during the shooting and other details. Appellant also explored the effect of Ms. Ervin's heavy cocaine use on her ability to perceive and remember the event, and her motive to curry favor with the government in order to obtain a lenient sentence in her pending criminal case. (Ms. Ervin testified pursuant to a cooperation agreement with the government.) As to Ms. Carter, appellant focused primarily on her lack of certainty and her failure to link him to the shooting, even tentatively, prior to trial. Ms. Carter admitted telling the grand jury that she did not recognize the person she saw running from the scene of the shooting and could not tell whether that person was appellant.

Although appellant did not take the stand, he called several witnesses in his defense. Sontia Lemon, appellant's wife, testified that they were at home on the morning of the shooting — they lived about eight or nine houses away on Kendall Street, which is behind, and parallel to, Corcoran Street — and that appellant ran outside to investigate when they heard the gunfire. She followed him to the scene of the shooting and saw him use his cell phone to summon the police. A second defense witness, Desmond Belt, claimed he too heard appellant telephone from the scene for help. Officer Bryant Collins, one of the officers who responded to the crime scene, confirmed that he took down appellant's name there. Another defense witness testified to having seen someone other than appellant running from the 1800 block of Corcoran Street after the shooting. The fleeing man appeared to have "something concealed on his right-hand side of his back." Finally, an analyst from a private testing facility testified that appellant was not the source of DNA found on cartridge casings and a piece of chewing gum recovered by the police from the vicinity of the shooting.

To bolster his defense of misidentification, appellant had hoped to present the expert opinion testimony of Dr. Lori Van Wallendael, an associate professor of cognitive psychology at the University of North Carolina. Appellant's proffer described Dr. Van Wallendael's qualifications and stated she would testify "about the factors present in this case that are known to adversely affect eyewitness perception, memory, and identification . . . includ[ing] stress, weapon focus, and deficient and unreliable identification procedures." The proffer stated that "[t]here is a broad consensus among the researchers in this area that (1) violence causing stress and emotional arousal at the time of an encounter negatively affects the accuracy of an eyewitness identification; and (2) the presence of a weapon at the time of the encounter leads to the phenomenon of `weapon focus' that in turn negatively affects the accuracy of an eyewitness identification." The proffer cited studies showing that many jurors misunderstand these effects and erroneously think, for example, that the violent nature of the crime would be likely to enhance rather than diminish the reliability of an eyewitness' memory. Further, according to the proffer, research on identification procedures had identified "certain best practices including (1) cautionary instructions that the culprit may or may not be in the lineup and that the police will continue to investigate even if no identification is made; (2) admonishing the eyewitnesses not to confer about their memories or identifications to prevent cross-contamination of evidence; (3) ensuring that the identification procedure is double blind — that is, that the law enforcement officer conducting the lineup or photo array does not know the identity of the suspect and that the eyewitness attempting an identification is told that the officer does not know the identity of the suspect; [and] (4) conducting the lineup or photo array sequentially or serially."

Appellant further proffered that Dr. Van Wallendael would testify about a phenomenon called "Unconscious Transference," defined as the "transfer of one person's identity to that of another person from a different setting, time, or context." In an unconscious transference, the proffer stated, "the witness misidentifies the familiar innocent person without having a conscious recollection of the previous exposure to him/her and without awareness of the true context of the previous exposure." Appellant asserted that this phenomenon could explain how "an innocent bystander at the crime scene [could be] falsely identified as being the criminal." Appellant's theory, for which he sought Dr. Van Wallendael's expert imprimatur, was that Ms. Edwards could have seen him at the scene immediately after the shooting and unconsciously confused him with the shooter he resembled.

The government moved in limine to exclude Dr. Van Wallendael's testimony, primarily on the grounds that "all of the points that the defense seeks to establish through an expert can be addressed through cross-examination of the witnesses themselves" and would not be "beyond the ken of the average layperson or juror." The government also disputed the relevance of Dr. Van Wallendael's proposed opinion testimony because the research on which she relied concerned the reliability of identifications of strangers, whereas the identifications of appellant in this case were by persons who knew him or (in the case of Ms. Edwards) "saw him on a second sighting."

The trial court granted the government's in limine motion without a hearing. In a brief oral ruling, the court excluded Dr. Van Wallendael's testimony on the ground that cross-examination would suffice to present the issues and enable the jury to assess the eyewitness identifications of appellant.

The trial lasted five days. The jury was out for approximately a day before it returned its verdict finding appellant guilty on all counts.

II.

Appellant contends — and appellee concedes — that the trial court erred in excluding the proffered expert testimony of Dr. Van Wallendael without properly conducting the particularized inquiry required by this court's decisions in Dyas v. United States and Benn v. United States (Benn II). In Benn II, we recognized that the defense "should be permitted to present expert testimony on the unreliability of eyewitness testimony in appropriate cases." We held that while the issue is committed to the trial court's discretion, the court must evaluate a particular defense proffer of such testimony in "the concrete setting" of the case under each of the criteria enumerated in Dyas. The determination of admissibility

must be case-specific, based on the proffered expert testimony, and must consider: (1) the current state of generally-accepted scientific research; (2) whether it is within the common knowledge of lay jurors; and (3) whether the testimony would assist the jury, taking into account the relevance and probative value of the proposed scientific evidence to the eyewitness identification in the case.[]

We cautioned in Benn II that while "there are certain cases where cross-examination may suffice to test the reliability of the identification made by an eyewitness . . . reliance on cross-examination cannot be automatic or reflexively adopted in lieu of proffered expert testimony" because, among other reasons, "the information that an expert can provide about research studies is different in nature and cannot be elicited from a lay witness during cross-examination."

As the government admits, the trial court's ruling in this case excluding Dr. Van Wallendael's proffered testimony did not satisfy Benn II's procedural requirements. Appellant's proffer described Dr. Van Wallendael's credentials, stated that experts had come to a general consensus that stress, weapon focus, and certain police identification procedures can negatively impact eyewitness accuracy, and cited studies showing that jurors tend to misunderstand the effects of those factors. Further, appellant proffered that Dr. Van Wallendael would testify about research on "unconscious transference," a phenomenon (if it exists) potentially applicable to Ms. Edwards's identification of appellant and rather plainly beyond the ken of lay jurors. The trial court did not evaluate appellant's proffer in light of the full three-part Dyas test, and its ruling that cross-examination of the eyewitnesses would render the expert testimony unnecessary was conclusory. It was not the product of an adequate weighing of "the efficacy of cross-examination . . . in the context of the particular case . . . against the value added to the jury of the proffered expert testimony."

Appellant asks us to remand this case for the court to conduct the necessary inquiry into whether the proffered testimony of Dr. Van Wallendael would satisfy the criteria for admission. Such a remand would be our normal response upon finding that a trial court excluded expert testimony proffered by the defense without properly evaluating it in accordance with Dyas and Benn II. However, the government argues that a remand is unnecessary in this case, because even if appellant should have been permitted to present Dr. Van Wallendael's proffered testimony to the jury, we can say with "fair assurance" that the erroneous exclusion of that testimony did not have a "substantial and injurious effect or influence in determining the jury's verdict" and hence was harmless. Appellant disagrees, and argues further that the government's burden is to show that the exclusion of Dr. Van Wallendael's testimony was harmless beyond a reasonable doubt, because it deprived him of his constitutional right to a meaningful opportunity to present his defense.

In order to address the parties' respective contentions, we are obliged to answer a threshold legal question: What is the test for determining when a trial court's erroneous exclusion of defense evidence violates the defendant's constitutional right to present a defense? We conclude that the correct test incorporates a familiar materiality requirement: There is no constitutional violation unless there exists a reasonable probability that the excluded evidence would have altered the jury's verdict in the defendant's favor.

Applying that test to the putatively erroneous exclusion of Dr. Van Wallendael's proffered testimony, we conclude that the error did not amount to a constitutional violation. We find it highly probable that the trial court's ruling did not influence the jury's verdict; we see no reasonable probability that the expert's testimony would have engendered an otherwise-nonexistent reasonable doubt of appellant's guilt in the jurors' minds. Similarly, we hold that the error was harmless under the test for non-constitutional error.

A. The Constitutional Right to Present a Defense

The Constitution guarantees a criminal defendant a meaningful opportunity to call witnesses in order to present a complete defense. The right is grounded in the Sixth Amendment's Compulsory Process Clause and is a component of the due process of law required by the Fifth and Fourteenth Amendments. But though the right is fundamental, it is not unlimited. In the first place, "[t]he accused does not have an unfettered right to offer testimony that is incompetent, privileged, or otherwise inadmissible under standard rules of evidence," and "well-established rules of evidence permit trial judges to exclude evidence if its probative value is outweighed by certain other factors such as unfair prejudice, confusion of the issues, or potential to mislead the jury." Evidentiary rules excluding evidence from criminal trials violate the constitutional right to present a defense only if they "infringe upon a weighty interest of the accused and are arbitrary or disproportionate to the purposes they are designed to serve."

A trial court's improper application of a constitutionally valid evidentiary rule to exclude relevant and competent defense evidence — the type of problem we confront in the present case — also may violate a defendant's constitutional right to present a defense. "Not every" such error will do so, however. Rulings excluding (or admitting) evidence ordinarily are within the ambit of the trial court's discretion and are subject to review only for abuse of that discretion under the Kotteakos standard of harmlessness. It is the "rare" case in which a trial court's "application of a rule of evidence is so erroneous and unfair" as to deprive a defendant of a meaningful opportunity to present a complete defense. "Only when the error deprives a defendant of a fair trial does it amount to a constitutional violation."

Thus, the Supreme Court has described the Sixth Amendment guarantee as "the accused's right to call witnesses whose testimony is `material and favorable to his defense.`" A materiality requirement is implicit in the right to a meaningful opportunity to present a complete defense, just as it is in other constitutional rights to obtain and utilize evidence — in particular, the Sixth Amendment right to compulsory process to secure defense witnesses and the due process right to the disclosure by the prosecution of exculpatory evidence in its possession. "At a minimum," this means "a defendant must demonstrate that the excluded evidence was important to his defense" in order to show that the error was of constitutional magnitude. The materiality inquiry therefore focuses on prejudice with respect to the fairness of the trial as a whole. Our cases have evinced that focus. "[F]or example," we stated in Clark v. United States, an error is of constitutional magnitude if it "wholly deprived" the defendant of "any opportunity" to present evidence on a "central issue" in the case. In contrast, we observed, "[i]f the issue is merely collateral, or where ample cross-examination has already been allowed or evidence admitted on a particular issue, trial court curtailment of the defendant's presentation does not implicate the defendant's Sixth Amendment rights" even if the excluded evidence is favorable to the defendant. Clark thus identified the two ends of the materiality spectrum, as it were. Subsequent decisions have held that the erroneous exclusion of defense evidence may violate the accused's constitutional right even if the accused is not "wholly deprived" of the opportunity to present a defense, so long as the excluded evidence is sufficiently critical to the defense.

The present case also involves a less-than-total interference with the presentation of defense evidence on a central issue at trial (namely, the reliability of the eyewitness identifications of appellant). But while our past cases envision the possibility of a constitutional challenge in these circumstances, they have not articulated a standard of materiality to use in judging when an erroneous evidentiary ruling resulting in exclusion of defense evidence rises to the level of a constitutional violation. The standard of materiality implicit in our cases is amorphous and unclear. It is desirable to clarify the standard so that we may resolve the question in this and future cases on a principled, rational, and proper basis.

The United States Court of Appeals for the Second Circuit has taken the lead in developing an objective materiality test "[t]o isolate those few situations" in which erroneous evidentiary rulings unconstitutionally restrict the accused's ability to present a defense. Borrowing the materiality standard set forth in United States v. Agurs for determining whether a prosecutorial failure to turn over exculpatory evidence to the defense violates due process (a so-called " Brady violation"), the Second Circuit holds that whether an erroneous exclusion of defense evidence violates the defendant's constitutional right to present a defense "depends upon whether `the omitted evidence [evaluated in the context of the entire record] creates a reasonable doubt that did not otherwise exist.'"

The test adopted by the Second Circuit has much to recommend it. It reflects the purpose of the constitutional guarantee to enable the defendant "to put before a jury evidence that might influence the determination of guilt." We think it appropriate, however, to clarify and rephrase the test in one important respect.

The test cannot literally be whether the appellate court finds that the omitted evidence creates a reasonable doubt of the defendant's guilt. Except in the clearest of cases, such a finding is the province of a jury (or other trier of fact) that considers the live testimony and other evidence placed before it — not of an appellate court that has only a cold paper record to assess. Normally the most an appellate court can say, with a greater or lesser degree of conviction, is that the excluded evidence could have given the jury a reason to doubt. Moreover, to state that materiality depends on whether the excluded evidence affected the outcome of the trial "does not establish a standard of materiality because it does not indicate what quantum of likelihood there must be that the undisclosed evidence would have affected the outcome."

The lack of a principled, workable materiality test is the problem that has plagued prior cases. For all the foregoing reasons, we think it necessary to reformulate the Second Circuit's test of unconstitutionality to reflect how likely it must be that the erroneously excluded evidence would have given the jury an otherwise-nonexistent reasonable doubt. Our choices are limited. As discussed above, it is well established that not every erroneous exclusion of evidence favorable to the defendant violates the defendant's constitutional right to present a defense. And in other contexts involving comparable claims of interference with the ability to present a defense, the Supreme Court has held that "a defendant cannot establish a constitutional violation simply by demonstrating that an alleged trial-related error could or might have affected the jury." Thus, the bare possibility of prejudice cannot suffice to show that the defendant actually was deprived of a meaningful opportunity to present a complete defense. At the opposite extreme, we cannot demand certainty that the excluded evidence would have generated a reasonable doubt — a simply unknowable proposition. Even to require a showing that the evidence more likely than not would have led to an acquittal is too rigorous and unrealistic, in our view. As the Supreme Court has held in addressing Brady violations and ineffective assistance of counsel claims, a preponderance of the evidence standard is inadequate to ensure that the defendant received a fair trial.

In the end, we are hard-pressed to come up with a better standard than the intermediate one required in the Brady context by the Supreme Court's post- Agurs decisions. Under the Court's reformulation of Agurs's materiality requirement, evidence withheld by the prosecution is material for due process purposes only "if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different." In other words, it must be reasonably probable (and not merely possible) that the jury would have harbored a reasonable doubt regarding the defendant's guilt if the evidence had not been suppressed. Yet the reasonable probability standard of materiality is not as demanding as a preponderance-of-the-evidence test:

The question is not whether the defendant would more likely than not have received a different verdict with the evidence, but whether in its absence he received a fair trial, understood as a trial resulting in a verdict worthy of confidence. A `reasonable probability' of a different result is accordingly shown when the government's evidentiary suppression `undermines confidence in the outcome of the trial.'[]

Notably, the reasonable probability standard is also the test of materiality the Supreme Court has employed for a violation of the Compulsory Process Clause, to which the constitutional right to present a defense is closely related. We think it most sensible to incorporate the reasonable probability standard into the Second Circuit's test. We hold, therefore, that whether an erroneous exclusion of defense evidence violates the defendant's constitutional right to present a defense depends upon whether there exists a reasonable probability that the omitted evidence, evaluated in the context of the entire record, would have led the jury to entertain a reasonable doubt that did not otherwise exist. Where such a reasonable probability exists, it makes sense to say the erroneous evidentiary ruling deprived the defendant of his constitutional right to a meaningful opportunity to present a complete defense, But if there is no reasonable probability that the outcome of the trial would have been different, we are prepared to conclude that the error was not so grave as to deprive the defendant of the meaningful opportunity guaranteed by the Sixth Amendment and the Due Process Clause — in other words, it was not constitutional error.

Although the proposition is not self-evident, the Supreme Court has stated that the reasonable probability standard of materiality "would recognize reversible constitutional error only when the harm to the defendant was greater than the harm sufficient for reversal under Kotteakos" for errors that do not implicate constitutional rights. If an evidentiary error in the exclusion of defense evidence is harmless under Kotteakos, it therefore cannot be material enough to amount to an error of constitutional magnitude. Conversely, if the error is sufficiently prejudicial to require reversal under Kotteakos, there seldom if ever will be any need for this court to consider on direct appeal whether a constitutional violation occurred. In other words, the Kotteakos analysis will be dispositive. B. The Probative Value of Dr. Van Wallendael's Proffered Testimony

In order to determine whether appellant was prejudiced by the exclusion of Dr. Van Wallendael's proffered testimony, we must assess its value to appellant's defense. The court's ruling did not wholly or even substantially deprive appellant of the opportunity to present evidence challenging the reliability of the eyewitnesses' identification testimony. Examining the expert's testimony not in a vacuum, but in the context of appellant's trial, we perceive no reasonable probability that it would have created an otherwise-nonexistent reasonable doubt of appellant's guilt in the minds of the jury. Rather, we find it highly probable that the trial court's presumptively erroneous exclusion of the testimony made no difference to the outcome of appellant's trial. We therefore conclude that the putative error was non-constitutional — it did not deprive appellant of a meaningful opportunity to present a complete defense — and harmless in Kotteakos terms.

We reach these conclusions because Dr. Van Wallendael's proffered testimony simply had too little to say about the identifications it was meant to undercut. Dr. Van Wallendael would have opined that stress, weapon focus, and deficient police procedures reduce the reliability of identifications made by crime victims and witnesses of unknown perpetrators, and that some misidentifications of strangers may be attributable to a phenomenon known as "unconscious transference." In the first place, this expert testimony would have been of scant relevance to the identifications of appellant by Ms. Ervin and Ms. Carter, whatever the reliability of those identifications, because they already knew appellant at the time of the shooting. There is no indication that their ability to recognize him as one of the shooters was impaired by any of the factors Dr. Van Wallendael would have discussed, or that an unconscious transference was at work in their cases.

The target of the expert's testimony would have been Ms. Edwards, the one identification witness to whom appellant was a stranger. Even so, Dr. Van Wallendael's expertise would have been unlikely to help the jury evaluate her identification. Ms. Edwards did not claim she saw appellant's face during or after the shooting; rather, she claimed to remember him as one of the men she saw approaching the car before the attack, at a time when she was at ease and not aware of any threat. Dr. Van Wallendael's testimony that stress and weapon focus may undermine an identification based on observations made under the influence of those factors therefore would have been beside the point. Appellant did not proffer that the circumstances of the shooting would have influenced Ms. Edwards's memory of preceding events. Similarly, so far as appears, Dr. Van Wallendael would not have identified any significant flaws in the procedure through which Ms. Edwards selected appellant's photograph.

Appellant hangs his hat on Dr. Van Wallendael's testimony regarding the "unconscious transference" phenomenon. He argues that this phenomenon would have explained how Ms. Edwards could have misidentified appellant as a result of having glimpsed him on the scene in the immediate aftermath of the shooting. But even assuming that, after a hearing, the trial court would have found the expert's testimony to be admissible under the criteria set forth in Dyas, the proffer stated only that an unconscious transference "can" occur. Among other things, the proffer said nothing about the prevalence of the phenomenon, the conditions under which it is likely to occur (or whether those conditions were present in this case), or how it may be recognized or ruled out. At most, in other words, it appears from the record that Dr. Van Wallendael might have testified that an unconscious transference was theoretically possible in this case, but not that it was likely or probable under circumstances resembling Ms. Edwards's experience. The possibility of an unconscious transference depended, moreover, on whether Ms. Edwards in fact saw appellant on the scene after the shooting. But Ms. Edwards did not remember seeing him there, and there is no evidence establishing that she did see him.

Further, Dr. Van Wallendael's testimony would not have undermined the central strength of the prosecution's case: the fact that Ms. Ervin and Ms. Edwards independently identified appellant as one of the shooters without having communicated with each other and without having learned that appellant was a suspect (not to mention the added fact that their identifications were corroborated by Ms. Carter). Viewing the witnesses' identifications in isolation, the jury potentially could have discounted them: Perhaps Ms. Ervin had a motive to frame appellant in order to obtain the benefits of her cooperation agreement, and perhaps Ms. Edwards's visceral reaction when appellant entered her store could be attributed to her unconscious memory of his presence on the scene after the shooting. (And perhaps Ms. Carter simply mistook the person she saw running from the scene for appellant because they did, in fact, resemble each other.) But the witnesses' identifications of appellant were mutually reinforcing, and that made the alternative explanations advanced by the defense less probable. The defense offered no explanation, other than sheer coincidence, for how Ms. Ervin, supposedly looking for someone to frame, happened to pick the one innocent person who, though innocent, would so distress Ms. Edwards in a chance meeting at her place of work some eight months after the shooting. Dr. Van Wallendael's testimony would not have made such an unlikely coincidence seem plausible.

Finally, the exclusion of the identification expert's testimony did not foreclose appellant from mounting a vigorous challenge to Ms. Edwards's identification of him. Appellant conducted "ample cross-examination" of Ms. Edwards on her ability to observe and identify the shooters. Notably, defense counsel cross-examined her about her post-traumatic reconstruction of events and her lack of a clear view during the attack, leaving little to be gained from an expert's pronouncements on stress and weapon focus. Counsel also interrogated Ms. Edwards about her adverse reactions to other men besides appellant who had dark complexions and dreadlocks. In closing, the defense emphasized the realistic possibility that Ms. Edwards had made a mistake in identifying appellant. Appellant was able to urge the jury to draw the commonsense inference that Ms. Edwards had mistaken appellant for one of the shooters based on her confused memories of the event and the similarity in appearance between the two men. To be sure, the jury did not hear about the abstract possibility of an unconscious transference, to which only an expert could have testified. But we are persuaded that testimony as limited as that which appellant proffered would not have made a difference.

The dissent contends that the prosecution's evidence was "weaker" in this case than it was in either Russell v. United States or Benn II, where we found reversible error in the trial court's failure to hold a Dyas hearing on expert identification testimony proffered by the defense. But the question is not whether the identification testimony was strong or weak; it is whether Dr. Van Wallendael's proffered testimony could have influenced the jury's decision in light of the facts in this particular case. In both Russell and Benn II, the proffered expert testimony was material to the credibility of each of the government eyewitnesses. In Russell, for example, expert opinion regarding weapons focus was highly relevant, because the victim in that case based his identification at least in part on observations he made during the armed robbery, even though he also happened to have seen his attacker prior to the crime. Similarly, in Benn II, the proffered expert testimony addressed the "unreliability [of] stranger-to-stranger eyewitness identifications," and the government's case hinged on five such identifications that were not independent: The witnesses were able to confer between their viewings of the photographic lineup, and they gave suspiciously similar responses, with three of them stating in identical language that they were "95% certain" they had identified the correct man. In contrast to the potentially material expert testimony proffered in Russell and Benn II, the proffered testimony in the present case would have been of scant relevance to the jury's evaluation of the identifications.

In sum, the probative value to appellant's defense of Dr. Van Wallendael's proffered testimony was slight at best. We see no reasonable probability that its putatively erroneous exclusion contributed to the verdict against appellant. Therefore, although the trial court did not evaluate the admissibility of the testimony under the Dyas criteria in accordance with our opinion in Benn II, we will not remand for the court to conduct that inquiry now. The error, if any, was non-constitutional and harmless.

III.

For the foregoing reasons, we affirm appellant's convictions. Because his convictions on two counts of possession of a firearm during a crime of violence ("PFCV") merge, we remand the case for the trial court to vacate one of them and determine whether resentencing on the remaining counts is necessary to effectuate its original sentencing plan. So ordered.