Int

0091-4169/20/11002-0181 THE JOURNAL OF CRIMINAL LAW & CRIMINOLOGY Vol. 110, No. 2

Copyright © 2020 by Lawrence Rosenthal Printed in U.S.A.

181

EYEWITNESS IDENTIFICATION AND THE

PROBLEMATICS OF BLACKSTONIAN

REFORM OF THE CRIMINAL LAW

LAWRENCE ROSENTHAL*

A substantial number of wrongful convictions are attributable to

inaccurate identifications of perpetrators, stemming from the difficulties that

eyewitnesses can experience in accurately perceiving and later recalling

faces. Many have argued that courts should employ prophylactic rules to

prevent the admission of unreliable identification evidence. Yet, most

jurisdictions continue to follow the deferential approach to the admission of

eyewitness identification evidence taken by the United States Supreme Court

in Manson v. Brathwaite. Commentators have universally condemned this

state of affairs.

This Article offers a departure from the existing commentary by taking

seriously the possibility that courts have good reason for their reluctance to

embrace prophylactic rules excluding evidence thought to present unduly

high risks of convicting the innocent.

The case for reform is rooted in Blackstone’s admonition that the law

should be wary of admitting evidence of guilt, preferring erroneous

acquittals to wrongful convictions. It is difficult, however, to construct a

Blackstonian case for the exclusion of evidence thought to be unduly likely

to produce wrongful convictions. Given our limited knowledge about the

error rates that inhere in most types of evidence, Blackstonian reform has no

ascertainable stopping point; excluding evidence that poses what is thought

to be an undue risk of wrongful conviction could result in the exclusion of

virtually all evidence of guilt. To illustrate the point, this Article considers

an issue on which the lower courts have split—the role of corroborative

* Professor of Law, Chapman University, Dale E. Fowler School of Law. The author has

benefitted from the incisive comments of Kevin Cole, Donald Kochan, Scott Howe, Eric Miller, Ric Simmons, Jonathan Witmer-Rich, and those who attended the presentation of

earlier drafts at Crimfest at Cardozo Law School and Southern California Criminal Justice

Roundtable held at the University of California, Irvine School of Law. The author is also

grateful to Rachel Cherny-Santos, Joseph Hernandez, Kaylee Sauvey, Jaime Traxler, Sherry Leysen and the staff of the Rinker Law Library for invaluable research assistance.

182 ROSENTHAL [Vol. 110

evidence in assessing the admissibility of an eyewitness’s identification.

Although Blackstonian prophylactic rules reject the consideration of

corroborative evidence, the reliability of most evidence cannot be assessed

in isolation. Reliability can usually be assessed only in the context of all

pertinent evidence. Thus, totality-of-the-circumstances tests, such as the

approach reflected in the Supreme Court’s decisions on eyewitness

identification, are about the best we can do—as prophylactic evidentiary

rules designed to reduce rates of wrongful conviction turn out to be deeply

problematic.

INTRODUCTION ..........................................................................183 I. MANSON V. BRATHWAITE AND THE LIMITED SCOPE OF

JUDICIAL GATEKEEPING IN THE ADMISSION OF

EYEWITNESS IDENTIFICATION EVIDENCE.................188 A. Manson’s Approach to Eyewitness Identification Evidence

.....................................................................................188 1. The Road to Manson ................................................189 2. The Holding in Manson ...........................................191 3. The Critique of Manson ...........................................192

B. State-Law Alternatives to Manson.................................194 1. Henderson and its Progeny ......................................194 2. Per se Exclusionary Rules ........................................197 3. Incremental Reforms ...............................................198

C. The Limited Significance of the State-Law Alternatives to Manson .....................................................................198 1. Henderson and its Progeny ......................................198 2. Per se Exclusionary Rules ........................................203 3. Incremental Reforms ...............................................204

II. THE DIFFICULTIES OF BLACKSTONIAN REFORM ...........206 A. The Effects of Prophylactic Reform ..............................207

1. The Costs and Benefits of More Rigorous Identification Protocols .................................................................207

2. The Problematic Case for Blackstonian Prophylactic Rules .......................................................................217

B. Assessing Reliability in Light of the Totality of Circumstances...............................................................229 1. The Role of Corroborative Evidence ........................230 2. Corroborated Identifications and the Right to a Fair

Trial ........................................................................236 3. The Difficulty of Abandoning Manson .....................240

CONCLUSION ..............................................................................241

2020] EYEWITNESS IDENTIFICATION 183

INTRODUCTION

More than a half-century ago, the Supreme Court wrote: “The vagaries

of eyewitness identification are well-known; the annals of criminal law are

rife with instances of mistaken identification.” 1 Time has done little to alter

this assessment; analyses of wrongful convictions continue to identify

inaccurate eyewitness identification as a leading cause of the conviction of

the innocent. 2

The reasons why eyewitness identifications lead to wrongful

convictions can be briefly summarized. 3 A large volume of research has

disclosed a substantial error rate in witnesses’ efforts to identify a suspect

accurately, stemming from the difficulties that witnesses can experience in

accurately perceiving and later recalling faces. 4 In particular, individuals

have a tendency to select the individual in a lineup or other identification

procedure who most resembles their recollection of the suspect, which injects

a substantial risk of error into identification evidence. 5 The research also

identifies a heightened risk of error when suggestive identification

procedures are employed—for example, when only the suspect or his picture

is shown to a witness (a “showup”), or the use of a lineup of individuals

(actual or photographic) in which only the suspect fits the witness’s previous

description of the perpetrator, or when witnesses receive instructions or

1 United States v. Wade, 388 U.S. 218, 228 (1967) (footnote omitted).

2 See, e.g., Russell D. Covey, Suspect Evidence and Coalmine Canaries, 55 AM. CRIM. L.

REV. 537, 546–48, 547 tbl.2 (2018); Brandon L. Garrett, Convicting the Innocent Redux, in

WRONGFUL CONVICTIONS AND THE DNA REVOLUTION: TWENTY-FIVE YEARS OF FREEING THE

INNOCENT 40, 45–46, 46 fig.3.3 (Daniel S. Medwed ed., 2017); Samuel Gross, What We Think, What We Know and What We Think We Know About False Convictions, 14 OHIO ST. J. CRIM.

L. 753, 769–73 (2017); Andrew M. Smith & Brian L. Cutler, Introduction: Identification

Procedures and Conviction of the Innocent, in REFORM OF EYEWITNESS IDENTIFICATION

PROCEDURES 3, 7–11 (Brian L. Cutler ed., 2013) [hereinafter REFORM OF EYEWITNESS IDENTIFICATION PROCEDURES].

3 A more detailed account of those aspects of the pertinent research most directly

applicable to this Article is found in Part II.A.1 below.

4 See, e.g., COMM. ON SCI. APPROACHES TO UNDERSTANDING & MAXIMIZING THE VALIDITY

AND RELIABILITY OF EYEWITNESS IDENTIFICATIONS IN LAW ENF’T & THE COURTS, IDENTIFYING

THE CULPRIT: ASSESSING EYEWITNESS IDENTIFICATION 45–101 (2014) [hereinafter IDENTIFYING THE CULPRIT]; ELIZABETH F. LOFTUS ET AL., EYEWITNESS TESTIMONY: CIVIL AND

CRIMINAL 13–78 (5th ed. 2013); BOAZ SANGERO, SAFETY FROM FALSE CONVICTIONS 182–85

(2016); Gary L. Wells, Eyewitness Identification, in 2 REFORMING CRIMINAL JUSTICE 259,

263–68 (Erik Luna ed., 2017).

5 See, e.g., Gary L. Wells & Eric P. Seelau, Eyewitness Identification: Psychological

Research and Legal Policy on Lineups, 1 PSYCHOL. PUB. POL’Y & L. 765, 768–69 (1995)

(summarizing research).

184 ROSENTHAL [Vol. 110

feedback encouraging them to make an identification. 6 Moreover, studies

have found that jurors have limited ability to assess the reliability of

eyewitness identifications and, instead, tend to over-believe eyewitnesses

and discount the risk of eyewitness error. 7 As one commentary put it:

Unlike accomplice witnesses, the typical eyewitness [to a crime] is a passerby who has

no motive to lie. Unlike circumstantial evidence, eyewitness testimony is directly

probative of guilt and frequently expressed with a high degree of certainty. Unlike

expert testimony, eyewitness testimony is immediately understood by even the most

confused, inattentive, or ignorant juror. And unlike many other kinds of evidence,

eyewitness testimony is rarely the subject of any cautionary instruction from the

judge . . . .8

In light of these problems, many have urged reforms to reduce the risk

of error created by potentially suggestive identification procedures, such as

training officers to avoid suggestive identification procedures; utilizing

double-blind procedures in which witnesses and lineup administrators are

unaware of the identity of the suspect; and instructing witnesses during

6 See, e.g., BRIAN L. CUTLER & STEVEN D. PENROD, MISTAKEN IDENTIFICATION: THE

EYEWITNESS, PSYCHOLOGY, AND THE LAW 113–36 (1995) (discussing risks of error related to lineup instruction bias, lineup construction, and lineup administrator bias); Steven E. Clark &

Ryan D. Godfrey, Eyewitness Identification and Innocence Risk, 16 PSYCHONOMIC BULL. &

REV. 22, 29–33 (2009) (discussing risks of error related to lineup construction, showup

identifications, lineup instructions, and simultaneous versus sequential lineups); David A. Sonenshein & Robin Nilon, Eyewitness Errors and Wrongful Convictions: Let’s Give Science

a Chance, 89 OR. L. REV. 263, 270–74 (2010) (discussing suggestiveness in showup

identifications, lineup construction, and administrator bias); Gary L. Wells & Deah S.

Quinlivan, Suggestive Eyewitness Identification Procedures and the Supreme Court’s Reliability Test in Light of Eyewitness Science: 30 Years Later, 33 LAW & HUM. BEHAV. 1, 6–

9 (2009) (discussing risks of error related to lineup instructions, lineup composition, showup

identifications, and administrator bias).

7 For discussions of the pertinent research, see, for example, CUTLER & PENROD, supra

note 6, at 181–96; DAN SIMON, IN DOUBT: THE PSYCHOLOGY OF THE CRIMINAL JUSTICE PROCESS 150–57 (2012); Steven E. Clark, Blackstone and the Balance of Eyewitness

Identification Evidence, 74 ALB. L. REV. 1105, 1147–52 (2010-11); Jennifer L. Devenport et

al., Eyewitness Identification Evidence: Evaluating Commonsense Evaluations, 3 PSYCHOL.

PUB. POL’Y & L. 338, 346–53 (1997); Richard S. Schmechel et al., Beyond the Ken? Testing Jurors’ Understanding of Eyewitness Reliability Evidence, 46 JURIMETRICS 177, 193–205

(2006); Dan Simon, On Juror Decisionmaking: An Empathic Inquiry, 15 ANN. REV. L & SOC.

SCI. 415, 419–20 (2019); Dan Simon, The Limited Diagnosticity of Criminal Trials, 64 VAND.

L. REV. 143, 152–60 (2011); and Gary L. Wells et al., Eyewitness Identification Procedures: Recommendations for Lineups and Photospreads, 22 LAW. & HUM. BEHAV. 603, 620–21

(1998).

8 Jed S. Rakoff & Elizabeth F. Loftus, The Intractability of Inaccurate Eyewitness

Identification, 147 DAEDALUS, Fall 2018, at 90, 91; cf. Perry v. New Hampshire, 565 U.S. 228, 260 (2012) (Sotomayor, J., dissenting) (“[J]urors find eyewitness evidence unusually

powerful and their ability to assess credibility is hindered by a witness’ false confidence in the

accuracy of his or her identification.”).

2020] EYEWITNESS IDENTIFICATION 185

lineups that the suspect or the suspect’s picture might not be present and that

they are free to make no identification. 9

In Manson v. Brathwaite, 10

the Supreme Court addressed the question

of whether an unreliable eyewitness identification can deprive a criminal

defendant of the constitutional right to a fair trial under the Due Process

Clause. The Court concluded that even when investigators utilize

unnecessarily suggestive procedures, an ensuing identification should not be

excluded from evidence absent a finding that it is unreliable in light of “the

totality of the circumstances”—rejecting “a strict exclusionary rule or new

standard of due process.” 11

Under this approach, “if the indicia of reliability

are strong enough to outweigh the corrupting effect of the police-arranged

suggestive circumstances, the identification evidence ordinarily will be

admitted, and the jury will ultimately determine its worth.” 12

Manson rarely

results in the exclusion of eyewitness identification evidence; a review of

federal cases available on the Westlaw database from the date Manson was

decided in 1977 until January of 2010, for example, found that identification

evidence was excluded under Manson in only 3.54% of cases, despite the use

of suggestive identification procedures in 57.10% of cases. 13

Manson reflects the predominant approach to judicial scrutiny of

eyewitness identification evidence; as one recent survey concluded: “[A]

large proportion of law enforcement agencies . . . have not made significant

reforms and most courts in the United States still use some version of the

Manson approach to dealing with eyewitness identification evidence.” 14

9 For discussions of recommended reforms along these lines, see, for example,

IDENTIFYING THE CULPRIT, supra note 4, at 106–07; Rakoff & Loftus, supra note 8, at 94–95; and Wells, supra note 4, at 277–78.

10 432 U.S. 98 (1977).

11 Id. at 113 (internal quotation and citation omitted).

12 Perry, 565 U.S. at 232.

13 Nicholas A. Kahn-Fogel, Manson and Its Progeny: An Empirical Examination of

American Eyewitness Law, 3 ALA. C.R. & C.L. L. REV. 175, 209–11, 211 tbl.1 (2012). The

prevalence of suggestive procedures fell only slightly over time. See id. at 220 (“[N]ear the middle of the time period, the probability of a case in the data set involving verifiable

suggestion was about 60%, but it was about 66% near the beginning of the time period and

only about 52% near the end.”). Moreover, despite the mounting social science evidence

illustrating the dangers of suggestive procedures, the willingness of courts to suppress evidence under Manson in cases involving suggestion actually decreased over time. See id.

(“[E]ven isolating only the 840 cases in which suggestion was evident, courts were

significantly less likely to suppress in-court identification evidence as time went on.”)

(footnote omitted).

14 Wells, supra note 4, at 276; see also Nicholas A. Kahn-Fogel, The Promises and Pitfalls

of State Eyewitness Identification Reforms, 104 KY. L.J. 99, 120 (2015-16) (“[T]he vast

majority of jurisdictions have followed Manson.”)

186 ROSENTHAL [Vol. 110

Commentators, however, have uniformly condemned Manson as inconsistent

with the large body of research that has emerged since that decision,

disclosing the perils of eyewitness identification. 15

The academic

commentary on eyewitness identification evidence contains nary a defense

of Manson. 16

The attacks on Manson as tolerating the admission of evidence thought

unduly likely to produce convictions of the innocent call to mind

Blackstone’s famous admonition: “[E]vidence of felony should be admitted

15 For helpful examples of the substantial volume of commentary criticizing Manson, see

JIM DWYER ET AL., ACTUAL INNOCENCE: FIVE DAYS TO EXECUTION AND OTHER DISPATCHES

FROM THE WRONGLY CONVICTED 74–75 (2000); BRANDON L. GARRETT, CONVICTING THE

INNOCENT: WHERE CRIMINAL PROSECUTIONS GO WRONG 53–54, 62–79 (2011); Clark, supra note 7, at 1135–42; Deborah Davis & Elizabeth F. Loftus, Inconsistencies between Law and

the Limits of Human Cognition, in MEMORY AND LAW 29, 49–53 (Lynn Nadel & Walter P.

Sinnott-Armstrong eds., 2012); Brandon L. Garrett, Eyewitnesses and Exclusion, 65 VAND. L.

REV. 449, 467–75 (2012); Sandra Guerra Thompson, Eyewitness Identifications and State Courts as Guardians Against Wrongful Conviction, 7 OHIO ST. J. CRIM. L. 603, 608–21

(2010); Aliza B. Kaplan & Janis C. Puracal, Who Could It Be Now? Challenging the

Reliability of First Time In-Court Identifications after State v. Henderson and State v. Lawson,

105 J. CRIM. L. & CRIMINOLOGY 947, 971–73 (2015); Timothy P. O’Toole & Giovanna Shay, Manson v. Brathwaite Revisited: Towards a New Rule of Decision for Due Process Challenges

to Eyewitness Identification Procedures, 41 VAL. U. L. REV. 109, 125–32 (2006); Benjamin

F. Rosenberg, Rethinking the Right to Due Process in Connection with Pretrial Identification

Procedures: An Analysis and a Proposal, 79 KY. L.J. 259, 275–97 (1990-91); Sonenshein & Nilon, supra note 6, at 274–78; Wells & Quinlivan, supra note 6, at 14–17; Richard A. Wise

et al., A Tripartite Solution to Eyewitness Error, 97 J. CRIM. L. & CRIMINOLOGY 807, 815–19

(2007).

16 About the only defense of Manson that can be found in the literature comes in a brief passage in a much longer article written by a prominent judge:

[C]ommentators have argued for, and some courts have even agreed to, broad restrictions on the

admissibility of eyewitness testimony . . . . Not only might it ultimately decrease the accuracy of

criminal verdicts – by taking fact-finding away from diverse juries and giving it to singular judges

– but it also impinges on the jury’s democratically grounded role as fact-finder. Nor are such

broad, judicially created rules of exclusion necessary. The Sixth Amendment dictates

confrontation rather than exclusion as the appropriate approach to eyewitness testimony. The

Confrontation Clause augments the jury’s role, and it is hardly up to judges to diminish it. Of

course . . . the Confrontation Clause excludes the out-of-court testimonial statements of witnesses

who do not testify at trial, unless that witness is “unavailable” and the defendant “had a prior

opportunity for cross-examination.” As a general matter, however, eyewitness testimony should

not be subject to a judge’s decision as to admissibility but should instead go through the adversary

process and be left to the jury’s determination of its value and weight.

J. Harvie Wilkinson III, In Defense of American Criminal Justice, 67 VAND. L. REV. 1099,

1161–62 (2014) (footnotes omitted). This argument fails to confront the evidence suggesting

that eyewitness identification evidence comes with a risk of error that is unlikely to be appreciated by jurors. If this is the case, the availability of a jury trial would not constitute an

adequate response to the risk of error injected into the criminal process by suggestive

identification procedures.

2020] EYEWITNESS IDENTIFICATION 187

cautiously: for the law holds, that it is better that ten guilty persons escape,

than that one innocent suffer.” 17

Indeed, commentators frequently invoke

Blackstone’s ratio to argue for more vigorous judicial policing of eyewitness

identification evidence. 18

Rather than adding to the chorus of criticism condemning Manson and

the prevailing approach to the admission of eyewitness identification

evidence, however, this Article offers a different perspective—by taking

seriously the possibility that courts have good reason for their reluctance to

embrace Blackstonian reform of criminal evidence law.

Part I reviews the development of the Manson test, as well as the

alternatives adopted by a handful of jurisdictions. It concludes that even

those jurisdictions that have embraced alternatives to Manson have not

achieved much in the way of meaningful reform.

Part II demonstrates that the virtues of Manson lie in the difficulties

presented by the alternatives, which involve the use of prophylactic rules

thought to minimize the risk of error. It is fiendishly difficult to know

whether such measures have benefits that exceed their costs. To illustrate the

point, Part II focuses on an issue on which the lower courts have split—the

role of corroborative evidence in assessing the admissibility of eyewitness

identification evidence. Blackstonian prophylaxis focuses on the procedures

employed to obtain an identification rather than on whether it is corroborated.

Any effort to assess the reliability of eyewitness identification evidence in

isolation, however, is deeply problematic. The reliability of evidence rarely

can be assessed in a vacuum; in the main, reliability is properly assessed in

light of the totality of the circumstances—including the available

corroborative evidence, or lack thereof. By disclaiming inquiry into

corroboration, accordingly, prophylactic rules would exclude a great deal of

reliable evidence.

The Blackstonian response to these difficulties rests on the view that we

should prefer false acquittals of the guilty to false convictions of the innocent.

Accordingly, if eyewitness identification evidence presents special risks to

the innocent, its admission should be viewed with special caution, even if

this will increase the rate at which the guilty go free. To be sure, a

Blackstonian preference for false acquittals over false convictions

undoubtedly underlies the heavy burden of proof that the prosecution must

17 4 WILLIAM BLACKSTONE, COMMENTARIES ON THE LAWS OF ENGLAND 352.

18 See, e.g., D. Michael Risinger, At What Cost? Blind Testing, Eyewitness Identification,

and What Can and Cannot Be Counted as a Cost of Reducing Information Available for Decision, 58 HOW. L.J. 333, 359 (2015) (“[T]he perceptive reader will hear an echo of the

Blackstone ratio . . . . [I]t is to be used as an approach to taking reformatory actions that will

improve the performance of a system-in-being at the margins.”).

188 ROSENTHAL [Vol. 110

shoulder in a criminal case. 19

It is a separate question, however, whether

additional safeguards are required when it comes to the admission of

eyewitness identification evidence. As Part II demonstrates, we know little

about the error rates that inhere in most types of evidence; the perils of

eyewitness identification evidence may not be unusual. There is,

accordingly, no identifiable stopping point for Blackstonian prophylactic

rules that exclude evidence giving rise to a risk of wrongful convictions. That

risk inheres in virtually all evidence. There is, therefore, little basis for

erecting special prophylactic exclusionary rules designed to screen out

evidence regarded as especially unreliable. Totality-of-the-circumstances

tests are about the best we can do.

I. MANSON V. BRATHWAITE AND THE LIMITED SCOPE OF

JUDICIAL GATEKEEPING IN THE ADMISSION OF

EYEWITNESS IDENTIFICATION EVIDENCE

The predominant view taken by courts as they assess the admissibility

of eyewitness identification evidence remains the highly deferential rule of

Manson v. Brathwaite. 20

There is, however, a minority view that demands a

more substantial showing of reliability before an eyewitness identification

can be placed before the jury.

A. MANSON’S APPROACH TO EYEWITNESS IDENTIFICATION

EVIDENCE

The traditional rule governing the admissibility of eyewitness

identification evidence was straightforward: “The overwhelming majority of

American courts have always treated the evidence question not as one of

admissibility but as one of credibility for the jury.” 21

On this view,

accordingly, judges played essentially no gatekeeping role—it was up to the

jury to assess the reliability of otherwise relevant eyewitness identification

evidence. 22

19 See, e.g., Speiser v. Randall, 357 U.S. 513, 525–26 (1958) (“Where one party has at

stake an interest of transcending value—as a criminal defendant his liberty—this margin of error is reduced as to him by the process of placing on the other party the burden of producing

a sufficiency of proof in the first instance, and of persuading the factfinder at the conclusion

of the trial of his guilt beyond a reasonable doubt.”). For a more general discussion of the basis

for the prosecution’s burden of proof beyond reasonable doubt in criminal cases, see In re Winship, 397 U.S. 358, 361–64 (1970).

20 432 U.S. 98 (1977).

21 Stovall v. Denno, 388 U.S. 293, 299–300 (1967) (citation omitted).

22 See, e.g., Perry v. New Hampshire, 565 U.S. 228, 245 (2012) (“[T]he jury, not the judge,

traditionally determines the reliability of evidence.”).

2020] EYEWITNESS IDENTIFICATION 189

1. The Road to Manson

The Supreme Court first broke with the traditional view in United States

v. Wade. 23

After observing that “[a] major factor contributing to the high

incidence of miscarriage of justice from mistaken identification has been the

degree of suggestion inherent in the manner in which the prosecution presents

the suspect to witnesses for pretrial identification,” 24

the Court worried that

“[t]he trial which might determine the accused’s fate may well not be that in

the courtroom but that at the pretrial confrontation . . . with little or no

effective appeal from the judgment there rendered by the witness—‘that’s

the man.’” 25

The Court therefore characterized the lineup at which Wade

was identified as the perpetrator of the crime as a “critical stage of the

prosecution” at which Wade was entitled to the assistance of counsel under

the Sixth Amendment. 26

Wade was something of a false start. Its practical significance was

limited five years later, when the Court held that the right to have counsel

present at an identification procedure did not extend to identifications

occurring before formal criminal charges are filed. 27

Subsequently, the Court

concluded that even for post-charging identifications, no right to counsel

attaches to the use of photographic identifications because of the ease with

which a photo array can be preserved for subsequent inspection. 28

But even

putting aside Wade’s limited scope, there is reason to doubt the efficacy of

its reliance on the presence of counsel as a vehicle for enhancing the

reliability of eyewitness identifications. At best, the presence of defense

counsel might assist the defense in identifying potential flaws in the process,

and perhaps deter overt misconduct. But even when defense counsel is

present, the absence of standards that require reliable identification

23 388 U.S. 218 (1967).

24 Id. at 228.

25 Id. at 235–36.

26 Id. at 236–38. In pertinent part, the Sixth Amendment provides: “In all criminal

prosecutions, the accused shall enjoy the right to have the Assistance of Counsel for his defence.” U.S. CONST. amend. VI. In a companion case decided the same day as Wade, the

Court held that the admission of an in-court identification without inquiry into whether it was

tainted by a prior lineup conducted in the absence of counsel in violation of the Sixth

Amendment was also constitutional error. See Gilbert v. California, 388 U.S. 263, 271–74 (1967).

27 See Kirby v. Illinois, 406 U.S. 682, 688–90 (1972) (plurality opinion); id. at 691

(Powell, J., concurring in the result).

28 See United States v. Ash, 413 U.S. 300, 313–21 (1973).

190 ROSENTHAL [Vol. 110

procedures means that the attorney would have few tools available to ensure

that reliable procedures are employed. 29

A more straightforward approach to reducing the risk of error created

by unreliable eyewitness identifications would involve regulating the process

of eyewitness identification itself to reduce the risk of error. Indeed, there is

a line of cases that points in that direction, beginning with the Supreme

Court’s decision—announced the same day as Wade—in Stovall v. Denno. 30

Stovall, the suspect in the case, was brought to the hospital room of a

stabbing victim in the custody of police officers—Mrs. Behrendt—who

identified him as the individual who had attacked her and killed her

husband. 31

While acknowledging that an identification of an alleged

perpetrator could be “so unnecessarily suggestive and conducive to

irreparable mistaken identification that he was denied due process of law[,]”

the Court concluded that “a claimed violation of due process of law in the

conduct of a confrontation depends on the totality of the circumstances

surrounding it, and the record in the present case reveals that the showing of

Stovall to Mrs. Behrendt in an immediate hospital confrontation was

imperative.” 32

After all, “[n]o one knew how long Mrs. Behrendt might

live.” 33

Stovall’s implication was that the use of evidence derived from an

unnecessarily suggestive identification procedure could deprive the accused

of due process of law. Subsequent cases, even as they rejected due process

claims, continued to leave open that possibility. 34

29 Cf. Donald A. Dripps, Miscarriages of Justice and the Constitution, 2 BUFF. CRIM. L.

REV. 635, 656 (1999) (“[T]he right to counsel approach does not focus on the underlying

problems with the reliability of the evidence. Giving the suspect a lawyer before a lineup does . . . what? The lawyer can testify as a witness about suggestiveness later on, but any

accurate recording of the session could achieve that much.”) (ellipsis in original and footnote

omitted); Garrett, supra note 15, at 466–67 (“[H]aving the right to a lawyer present at a lineup

is not a significant protection . . . . At best, it may discourage police from making any obviously suggestive cues during the lineup itself, though with the cost of potentially turning

the lawyer into a trial witness disqualified from further representation.”); Louis Michael

Seidman, Soldiers, Martyrs, and Criminals: Utilitarian Theory and the Problem of Crime

Control, 94 YALE L.J. 315, 328 (1984) (“The presence of an attorney at certain identification procedures enhances the illusion of accuracy while doing little to remedy the problems

identified by perceptual psychologists.”).

30 388 U.S. 293 (1967).

31 Id. at 295.

32 Id. at 302.

33 Id. (quoting United States ex rel. Stovall v. Denno, 355 F.2d 731, 735 (2d Cir. 1966),

aff’d sub nom. Stovall v. Denno, 388 U.S. 293 (1967)).

34 See, e.g., Simmons v. United States, 390 U.S. 377, 384–85 (1968) (upholding use of

identifications of photographs of suspects by observing that “it is not suggested that it was

unnecessary for the FBI to resort to photographic identification” since “the perpetrators were

2020] EYEWITNESS IDENTIFICATION 191

Stovall’s implication became a square holding in Foster v. California. 35

In that case, the sole witness to a robbery viewed Foster in a three-person

lineup in which Foster was much taller than the fillers and wore a jacket

similar to the one the witness had seen the robber wearing, yet the witness

was unable to make a positive identification then, or at a subsequent showup

that only included Foster; but she finally identified Foster at a second lineup,

in which Foster was the only individual who had also appeared in the first

one. 36

Citing Stovall, the Court wrote: “[t]he suggestive elements in this

identification procedure made it all but inevitable that [the witness] would

identify [Foster] whether or not he was in fact ‘the man.’ . . . This procedure

so undermined the reliability of the eyewitness identification as to violate due

process.” 37

Foster established that a conviction resting on evidence derived from

unnecessarily suggestive identifications could deprive an accused of the right

to a fair trial under the Due Process Clause. The limits of this holding,

however, became clear in Manson.

2. The Holding in Manson

The facts of Manson were straightforward. After an undercover police

officer provided narcotics officers with a description of the individual who

had just sold him heroin, a photograph was left on the undercover officer’s

desk, and two days later the officer identified the photograph as depicting the

seller. 38

While the majority conceded that “the procedure in the instant case

was suggestive because only one photograph was used and

unnecessary because there was no emergency or exigent circumstance,” 39

the

Court rejected the view that evidence obtained through unnecessarily

suggestive procedures “automatically is to be excluded.” 40

The Court

reasoned that “[t]he standard, after all, is that of fairness required by the Due

Process Clause,” and, therefore, courts must consider “the totality of the

circumstances.” 41

For the Court, “reliability is the linchpin in determining

still at large,” and adding that “there was in the circumstances of this case little chance that

the procedure utilized led to misidentification of Simmons”).

35 394 U.S. 440 (1969).

36 Id. at 441–42.

37 Id. at 443.

38 Manson v. Brathwaite, 432 U.S. 98, 99–101 (1977).

39 Id. at 109 (citations and internal quotations omitted).

40 Id. at 113.

41 Id. (citation and internal quotations omitted).

192 ROSENTHAL [Vol. 110

the admissibility of identification testimony” 42

and, when assessing

reliability, the relevant factors

include the opportunity of the witness to view the criminal at the time of the crime, the

witness’ degree of attention, the accuracy of his prior description of the criminal, the

level of certainty demonstrated at the confrontation, and the time between the crime

and the confrontation. Against these factors is to be weighed the corrupting effect of

the suggestive identification itself.43

Applying those factors, the Court wrote that the identification enjoyed

adequate indicia of reliability, 44

adding that “we cannot say that under all the

circumstances of this case there is ‘a very substantial likelihood of irreparable

misidentification.’” 45

3. The Critique of Manson

Even though Manson’s requirement of reliability review for proffered

eyewitness identification evidence produced by potentially suggestive

procedures represents a break from the general rule that the reliability of

otherwise relevant evidence should be assessed by the trier of fact, Manson

has been subject to fierce criticism for failing to erect an adequate barrier to

the admission of unreliable identifications.

Perhaps the primary ground of attack is that Manson’s account of

reliability is inconsistent with the growing body of research studying the

reliability of eyewitness identifications. For example, commentators have

argued that Manson’s direction to assess reliability in light of the witness’s

initial description of the suspect is inconsistent with subsequent research

finding little correlation between the accuracy of a suspect’s initial

42 Id. at 114. To support this novel requirement of judicial inquiry into the reliability of

proffered evidence in the face of official suggestion, the Court cited Neil v. Biggers, 409 U.S.

188 (1972). Manson, 432 U.S. at 114. In Neil, the Court had upheld the use of a showup

identification of the suspect by concluding that the identification “was reliable even though the confrontation procedure was suggestive.” See Neil, 409 U.S. at 199–201. As the Court

acknowledged in Manson, however, Neil’s precise holding did not control since, in that case,

“the challenged procedure occurred pre-Stovall and that a strict rule would make little sense

with regard to a confrontation that preceded the Court’s first indication that a suggestive procedure might lead to the exclusion of evidence. One perhaps might argue that, by

implication, the Court suggested that a different rule could apply post-Stovall.” Manson, 432

U.S. at 107 (citation omitted).

43 Manson, 432 U.S. at 114.

44 Id. at 114–16.

45 Id. at 116 (quoting Simmons v. United States, 390 U.S. 377, 384 (1968)). Subsequently the Court held that the Due Process Clause does not entitle an accused to a hearing outside the

presence of the jury on the admissibility of eyewitness identification evidence. See Watkins v.

Sowders, 449 U.S. 341, 346–49 (1981).

2020] EYEWITNESS IDENTIFICATION 193

description and the accuracy of a subsequent identification. 46

Similarly,

Manson is attacked because subsequent research has disclosed that other

factors Manson relied upon to establish reliability, such as a witness’s

opportunity to view the perpetrator, degree of certainty, and degree of

attention, are often difficult to assess and are themselves subject to

manipulation by suggestive identification procedures. 47

Critics also charge

that Manson’s multi-factor test permits courts to admit eyewitness

identification evidence even when critical factors identified in the research

suggest a heightened risk of unreliability, such as when the witness had a

poor opportunity to view the perpetrator or made an identification after

substantial time has passed. 48

Some add that Manson ignores research

disclosing the special risks of cross-racial identifications, 49

the risk of error

created by the tendency of witnesses to focus on the presence of weapons, 50

the stress of witnessing a violent crime, 51

or the special risks in identifications

by child or juvenile witnesses. 52

Beyond Manson’s claimed inconsistency with the pertinent social

science research, critics argue that Manson’s focus on reliability encourages

courts to overlook unnecessarily suggestive identification procedures if they

are convinced the defendant is guilty. 53

And critics also contend that the

reliability review contemplated by Manson is insufficiently protective of the

innocent because it permits unreliable identifications to be admitted in

evidence as long as the suggestion can be characterized as somehow

necessary to the investigation. 54

46 See, e.g., GARRETT, supra note 15, at 69–70; Sonenshein & Nilon, supra note 6, at 276.

47 See, e.g., GARRETT, supra note 15, at 71–72; Thompson, supra note 15, at 610–11, 613;

Kahn-Fogel, supra note 14, at 115–17; Rosenberg, supra note 15, at 276–79; Sonenshein &

Nilon, supra note 6, at 275–76, 277; Wise et al., supra note 15, at 815–16, 817–18.

48 See, e.g., GARRETT, supra note 15, at 70–71, 72; Sonenshein & Nilon, supra note 6, at 277–78.

49 See, e.g., GARRETT, supra note 15, at 72–74; Thompson, supra note 15, at 605–06, 613;

Kahn-Fogel, supra note 14, at 117; Rosenberg, supra note 15, at 279.

50 See, e.g., Thompson, supra note 15, at 613; Kahn-Fogel, supra note 14, at 117;

Rosenberg, supra note 15, at 279–80.

51 See, e.g., Thompson, supra note 15, at 616, 618; Kahn-Fogel, supra note 14, at 117; Rosenberg, supra note 15, at 278.

52 See, e.g., GARRETT, supra note 15, at 75–77.

53 See, e.g., Brandon L. Garrett, Innocence, Harmless Error, and Federal Wrongful

Conviction Law, 2005 WIS. L. REV. 35, 83–85.

54 See, e.g., Thompson, supra note 15, at 610, 614–16. Although Manson contained no

square holding on this precise point, the Court subsequently held that a defendant may not challenge the admission of eyewitness identification evidence without establishing

unnecessarily suggestive police conduct. See Perry v. New Hampshire, 565 U.S. 228, 240–48

(2012).

194 ROSENTHAL [Vol. 110

B. STATE-LAW ALTERNATIVES TO MANSON

Despite the manifold attacks on Manson, as we have seen, its approach

to the admission of eyewitness identification evidence continues to be

followed by the vast majority of jurisdictions. 55

The courts in some

jurisdictions, however, have taken a different road. 56

1. Henderson and its Progeny

Likely the leading example of an alternative to Manson is State v.

Henderson. 57

In that case, the New Jersey Supreme Court reviewed research

demonstrating the large risk of error in eyewitness identifications, the

increased risk of error that may result from the manner in which identification

procedures are conducted (“systems variables”), and the circumstances under

which the witness viewed the subject (“estimator variables”), as well as the

risk that the jurors do not appreciate the risks of misidentification. 58

While

acknowledging that “[w]e have no authority, of course, to modify Manson,”

the New Jersey Supreme Court rested its decision on the due process rights

guaranteed by the state constitution. 59

The court held that when a defendant

can discharge “the initial burden of showing some evidence of

suggestiveness that could lead to a mistaken identification,” 60

the burden of

proof shifts to the prosecution “to show that the proffered eyewitness

55 See supra note 14 and accompanying text.

56 Some jurisdictions that have undertaken legislative or administrative reforms in

identification procedures that do not call for the exclusion of evidence obtained in violation of recommended procedures. For a helpful discussion of these reforms, see Keith A. Findley,

Implementing the Lessons from Wrongful Convictions: An Empirical Analysis of Eyewitness

Identification Reform Strategies, 81 MO. L. REV. 377, 411–51 (2016).

57 27 A.3d 872 (N.J. 2011).

58 Id. at 894–912.

59 Id. at 919, n.10.

60 Id. at 920. The court added that “concerns about estimator variables alone cannot trigger

a pretrial hearing; only system variables would.” Id. at 922. To support this conclusion, the

court reasoned that “eyewitness identification evidence will likely not be ruled inadmissible

at pretrial hearings solely on account of estimator variables,” “courts cannot affect estimator variables; by definition, they relate to matters outside the control of law enforcement,”

“suggestive behavior can distort various other factors that are weighed in assessing reliability.

That warrants a greater pretrial focus on system variables,” and, finally, that “to allow hearings

in the majority of identification cases might overwhelm the system with little resulting benefit.” Id. at 923. In a companion case involving a suggestive identification involving no

official conduct, the court “ma[d]e one modification to Henderson in applying it to cases

where there is no police action: we require a higher, initial threshold of suggestiveness to

trigger a hearing, namely, some evidence of highly suggestive circumstances as opposed to simply suggestive conduct.” State v. Chen, 27 A.3d 930, 942–43 (N.J. 2011). In this sense,

the court apparently prioritized regulating investigative conduct over policing the reliability

of evidence. As we will see, policing the reliability of evidence is a fraught enterprise.

2020] EYEWITNESS IDENTIFICATION 195

identification is reliable—accounting for system and estimator

variables . . . .” 61

The court identified the relevant “system” variables to be

such considerations as whether there was blind administration of the

identification procedure, pre-identification instructions, lineup construction,

feedback from investigators, the witness’s degree of confidence, the number

of viewings, whether suspects were viewed simultaneously or sequentially,

whether the witness produced a composite sketch of the suspect prior to the

identification procedure, and whether a showup identification took place. 62

The court identified the relevant “estimator” variables as whether the witness

was under stress at the time of the underlying events, whether a visible

weapon was used, the duration of the witness’s opportunity to view the

perpetrator during the underlying events, distance and lighting at that time,

the witness’s age and level of intoxication during the underlying events, the

perpetrator’s age, the perpetrator’s characteristics, the time elapsed between

the underlying events and the identification, whether cross-racial

identification is involved, whether multiple witnesses could have

contaminated an identification, and the speed of the identification. 63

61 Henderson, 27 A.3d at 920. Although the pertinent federal and state constitutional

provisions employ somewhat different formulations, there is no apparent textual basis that

explains the New Jersey Supreme Court’s decision to impose more stringent requirements

than are found in Manson. Compare N.J. CONST. art. I, § 1 (“All persons are by nature free and independent, and have certain natural and unalienable rights, among which are those of

enjoying and defending life and liberty, of acquiring, possessing, and protecting property, and

of pursuing and obtaining safety and happiness.”), with U.S. CONST. amend. XIV, § 1 (“No

State shall . . . deprive any person of life, liberty, or property, without due process of law.”). Moreover, in Henderson, the court relied on its supervisory power over the administration of

justice to require that law enforcement officers “make a full record—written or otherwise—

of the witness’ statement of confidence once an identification is made,” 27 A.3d at 900, and

that “police officers ask witnesses, as part of the identification process, questions designed to elicit (a) whether the witness has spoken with anyone about the identification and, if so, (b)

what was discussed. That information should be recorded and disclosed to defendants.” Id. at

909. It is accordingly difficult to read Henderson as being rooted in the peculiarities of New

Jersey law. Instead, it seems to reflect the court’s disagreement with Manson’s approach.

62 Henderson, 27 A.3d at 896–903.

63 Id. at 904–10. The court also directed that “enhanced instructions be given to guide

juries about the various factors that may affect the reliability of an identification in a particular

case.” Id. at 924. In another decision announced on the same day, the court added: “[W]e make

one modification to Henderson in applying it to cases where there is no police action: we require a higher, initial threshold of suggestiveness to trigger a hearing, namely, some

evidence of highly suggestive circumstances as opposed to simply suggestive conduct.” Chen,

27 A.3d at 942–43.

196 ROSENTHAL [Vol. 110

Henderson has received fulsome praise from commentators. 64

It has

influenced other jurisdictions as well. In State v. Lawson, 65

for example, the

Oregon Supreme Court, relying in significant part on Henderson, 66

interpreted its rules of evidence to require that when there is evidence of

police suggestion, “the state—as the proponent of the identification—must

establish by a preponderance of the evidence that the identification was based

on a permissible basis rather than an impermissible one, such as suggestive

police procedures.” 67

Even when the state discharges this burden, a court

may nevertheless find that “the probative value of the evidence is

substantially outweighed by the danger of unfair prejudice, confusion of the

issues, misleading the jury, or by considerations of undue delay or needless

presentation of cumulative evidence.” 68

Subsequently, the Alaska Supreme

Court, relying on its supervisory power over the administration of criminal

justice as well as Henderson, held that when a defendant adduces evidence

of official suggestiveness, a trial court must hold an evidentiary hearing and

consider evidence on all relevant system and estimator variables. 69

Similarly,

the Connecticut Supreme Court adopted Henderson’s holding as an

interpretation of the Connecticut Constitution’s due process clause. 70

64 See, e.g., Keith A. Findley, Judicial Gatekeeping of Suspect Evidence: Due Process and

Evidentiary Rules in the Age of Innocence, 47 GA. L. REV. 723, 738–51 (2013); GARRETT,

supra note 15, at 493–94; Barry Scheck, Four Reforms for the Twenty-First Century, 96

JUDICATURE 323, 333–36 (2013); Wells, supra note 4, at 275; Robert Couch, Comment, A Model for Fixing Identification Evidence after Perry v. New Hampshire, 111 MICH. L. REV.

1535, 1542–46 (2013); Amy D. Trenary, Comment, State v. Henderson: A Model for

Admitting Eyewitness Identification Testimony, 84 U. COLO. L. REV. 1257, 1295–300 (2013);

Dana Walsh, Note, The Dangers of Eyewitness Identification: A Call for Greater State Involvement To Ensure Fundamental Fairness, 36 B.C. INT’L & COMP. L. REV. 1415, 1449–

53 (2013).

65 291 P.3d 673 (Or. 2012).

66 Id. at 685 n.3.

67 Id. at 693. The court based its holding on the requirements in the Oregon Rules of

Evidence that lay opinion testimony be rationally based on the witness’s perceptions, helpful to the trier of fact, and not unfairly prejudicial, reasoning that suggestive identifications are in

tension with these requirements. Id. at 693–95. Given that the Manson Court regarded its

approach as placing only reliable evidence before the jury, like Henderson, it is difficult to

understand Lawson as based on some peculiarity of state law as opposed to a substantive disagreement with Manson.

68 Id. at 697. The court added that expert testimony is appropriate to inform the trier of

fact about the variables that can affect the reliability of identification testimony. Id. at 695–

96.

69 See Young v. State, 374 P.3d 395, 412–28 (Alaska 2016).

70 See State v. Harris, 191 A.3d 119, 133–47 (Conn. 2018). With respect to the estimator variables to be employed in assessing reliability, the court indicated that “the trial court should

consider the eight estimator variables that this court identified in State v. Guilbert, [49 A.3d

705 (Conn. 2012)], which overlap considerably with the estimator variables identified in

2020] EYEWITNESS IDENTIFICATION 197

2. Per se Exclusionary Rules

Some jurisdictions have adopted approaches even more skeptical of

eyewitness identification evidence. The highest courts in Massachusetts and

New York, for example, have embraced a per se rule of exclusion of any

identification resulting from the use an unnecessarily suggestive

identification procedure under the due process clauses of their state

constitutions. 71

The Wisconsin Supreme Court once held that because

showup identifications are suggestive, evidence resulting from them should

be suppressed absent a showing that the procedure was necessary, although

it later retreated from that view. 72

A number of commentators have also

embraced robust rules of exclusion when identifications are made through

unnecessarily suggestive procedures. 73

Henderson. As we recognized in Guilbert, these variables are neither ‘exclusive’ nor ‘frozen in time.’” Id. at 144 (footnote omitted) (citations omitted).

71 See Commonwealth v. Johnson, 650 N.E.2d 1257, 1259–65 (Mass. 1995); People v.

Adams, 423 N.E.2d 379, 383–84 (N.Y. 1981).

72 See State v. DuBose, 699 N.W.2d 582, 591–99 (Wis. 2005), overruled by State v.

Roberson, 935 N.W.2d 813 (Wis. 2019).

73 See, e.g., Dripps, supra note 29, at 657–64 (requiring exclusion unless identification procedures comport with best practices and permitting defense counsel to participate when the

suspect is in custody); Susan R. Klein, Identifying and (Re)Formulating Prophylactic Rules,

Safe Harbors, and Incidental Rights in Constitutional Criminal Procedure, 99 MICH. L. REV.

1030, 1064–65 (2001) (advocating prophylactic rules requiring the use of non-suggestive procedures for identifications); Margery Malkin Koosed, Reforming Eyewitness Identification

Law and Practices To Protect the Innocent, 42 CREIGHTON L. REV. 595, 624–31 (2009)

(advocating exclusion of all identifications obtained through suggestive procedures); Robert

P. Mosteller, The Duke Lacrosse Case, Innocence, and False Identifications: A Fundamental Failure to “Do Justice”, 76 FORDHAM L. REV. 1337, 1408–12 (2007) (advocating per se

exclusion of identifications obtained without complying with procedures designed to

minimize suggestiveness); O’Toole & Shay, supra note 15, at 136–44 (advocating use of

prophylactic guidelines reflecting best practices identified through empirical research); Chase T. Rogers, Putting Meat on Constitutional Bones: The Authority of State Courts to Craft

Prophylactic Rules under the Federal Constitution, 98 B.U. L. REV. 541, 573–77 (2018)

(advocating prophylactic rules governing the admissibility of in-court identifications of the

defendant); Rosenberg, supra note 15, at 297–314 (advocating per se exclusion of identifications obtained through unnecessarily suggestive procedures and exclusion of

identifications obtained through other suggestive procedures unless they satisfy a proposed

standard of probativeness, and guaranteeing the defense a right to expert testimony on

eyewitness identification); Wise et al., supra note 15, at 842–65, 868–71 (advocating use of prophylactic guidelines reflecting best practices identified through empirical research);

Suedabeh Walker, Comment, Drawing on Daubert: Bringing Reliability to the Forefront in

the Admissibility of Eyewitness Identification Evidence, 62 EMORY L.J. 1205, 1234–41 (2013)

(advocating heightened reliability screening for identification evidence); see also GARRETT, supra note 15, at 488–91 (advocating exclusion of in-court identifications of the defendant as

impermissibly suggestive); Evan J. Mandery, Due Process Considerations of In-Court

Identifications, 60 ALB. L. REV. 389, 420–21 (1996).

198 ROSENTHAL [Vol. 110

3. Incremental Reforms

Other states have undertaken more modest reforms. The Utah Supreme

Court has interpreted the due process clause of its state constitution to require

that reliability be assessed not merely in light of the factors mentioned in

Manson, but instead by reference to all factors that have been shown to affect

the reliability of an identification. 74

The Kansas Supreme Court subsequently

followed Utah’s approach. 75

The Maine Supreme Court has held that even

when there has been no showing of unnecessary official suggestion, Maine’s

rules of evidence require a court to determine whether an identification is

reliable. 76

By statute, North Carolina regulates identification procedures to

minimize the risk of suggestiveness—requiring, in particular, that lineups be

conducted by blind administrators when practicable, the sequential

presentation of suspects, and the use of fillers that match the suspect’s

description. 77

Additionally, the statute authorizes courts to suppress

identification evidence obtained without complying with the statutory

requirements. 78

Ohio has enacted a more limited statute requiring the use of

blind administrators when practicable, and permitting courts to suppress

identification evidence obtained in violation of the statute. 79

C. THE LIMITED SIGNIFICANCE OF THE STATE-LAW

ALTERNATIVES TO MANSON

It is unclear whether the limitations on the admissibility of eyewitness

identification evidence that various states have adopted have much in the way

of teeth. Indeed, there is reason to believe that they will rarely result in the

exclusion of evidence.

1. Henderson and its Progeny

Consider again Henderson. Even as it adopted what purported to be a

more robust rule of exclusion than Manson, Henderson cautioned, “[I]n the

vast majority of cases, identification evidence will likely be presented to the

jury. The threshold for suppression remains high.” 80

Then, citing Manson,

the court added that “the ultimate burden remains on the defendant to prove

74 See State v. Ramirez, 817 P.2d 774, 779–81 (Utah 1991).

75 See State v. Hunt, 69 P.3d 571, 573–77 (Kan. 2016).

76 See State v. Davis, 191 A.3d 1147, 1155–57 (Me. 2018).

77 Id.

78 See N.C. GEN STAT. ANN. § 15A-284.52 (West 2016).

79 See OHIO REV. CODE ANN. § 2933.83 (West 2016).

80 State v. Henderson, 27 A.3d 872, 928 (N.J. 2011).

2020] EYEWITNESS IDENTIFICATION 199

a very substantial likelihood of irreparable misidentification.” 81

That sounds

a great deal like Manson, which rejected suppression absent “a very

substantial likelihood of irreparable misidentification.” 82

That is reason

enough to question the practical significance of Henderson. Indeed, the facts

and actual outcome of the Henderson litigation illustrate its limited

significance.

Start with the facts. After Rodney Harper and James Womble spent

New Year’s Eve drinking wine and champagne and smoking crack cocaine

in the apartment of Womble’s girlfriend, Harper left the apartment around

10:15 P.M., returning around 2:00 A.M., and soon after his return, two men

forced their way inside. 83

Womble knew one of them, George Clark, who

took Harper into another room while the stranger trained a gun on Womble,

telling him, “Don’t move, stay right here, you’re not involved in this.” 84

Womble “remained with the stranger in a small, narrow, dark hallway,” and

later “testified that he ‘got a look at’ the stranger, but not ‘a real good

look.’” 85

Womble then overheard an argument between Clark and Harper in

the other room, followed by a gunshot. 86

Womble walked into the room, saw

Clark holding a handgun, and “[a]s Clark left, he warned Womble, ‘Don’t rat

me out, I know where you live.’” 87

On January 11, Detective Luis Ruiz and Investigator Randall MacNair

interviewed Womble, who told them that while he was in the apartment, he

heard two gunshots outside, and then found Harper slumped over his car in a

nearby parking lot, where Harper told Womble that he had been shot by two

men he did not know. 88

The next day, the officers confronted Womble with

inconsistencies in his story, and Womble claimed that the officers threatened

to charge him in connection with the murder. 89

Womble “admitted that he

lied at first because he did not want to ‘rat’ out anyone and ‘didn’t want to

get involved’ out of fear of retaliation against his elderly father.” 90

Womble

81 Id. at 920.

82 Manson v. Brathwaite, 432 U.S. 98, 116 (1977) (internal quotation and citation omitted).

83 Henderson, 27 A.3d at 879.

84 Id.

85 Id.

86 Id.

87 Id. In particular, the evidence at the ensuing trial was, “Womble smoked two bags of

crack cocaine with his girlfriend in the hours before the shooting; the two also consumed one bottle of champagne and one bottle of wine; the lighting was ‘pretty dark’ in the hallway where

Womble and [Henderson] interacted . . . .” Id. at 882.

88 Id.

89 Id.

90 Id.

200 ROSENTHAL [Vol. 110

then “led the investigators to Clark,” who ultimately gave a statement and

identified his confederate as Larry Henderson. 91

On January 14, Womble viewed a photo array at the prosecutor’s office

conducted by Detective Thomas Weber pursuant to guidelines issued by the

New Jersey Attorney General providing that “primary investigators should

not administer photo or live lineup identification procedures ‘to ensure that

inadvertent verbal cues or body language do not impact on a witness.’” 92

The

array consisted of seven “filler” photos and one photo of Henderson. 93

After viewing the array, Womble eliminated six photos, but “said he

‘wasn’t 100 percent sure of the final two pictures.’” 94

Detective Weber left

the room and told Inspector MacNair and Detective Ruiz that Womble could

not make a final identification, and at that point, MacNair and Ruiz entered

the interview room believing, according to MacNair’s subsequent testimony,

that Womble “was holding back—as he had earlier in the investigation—

based on fear.” 95

Inspector MacNair testified that he “just told him to focus,

to calm down, to relax and that any type of protection that [he] would need,

any threats against [him] would be put to rest by the Police Department,” and

Detective Ruiz told Womble to “just do what you have to do, and we’ll be

out of here.” 96

According to MacNair’s testimony, at that point Womble said

he “could make [an] identification.” 97

MacNair and Ruiz then left the

interview room, and Weber returned and again displayed the photos to

Womble sequentially, but this time, when Womble saw Henderson’s photo,

he made an identification. 98

Womble never recanted that identification,

although he later testified that “he felt as though Detective Weber was

‘nudging’ him to choose [Henderson]’s photo, and ‘that there was pressure’

to make a choice.” 99

91 Id. at 879–80. Specifically, Womble “advised that the man who shot Harper was known

as ‘Bubbles,’ and he indicated where ‘Bubbles’ lived. A database search of this address

generated a photograph of co-defendant Clark, whom Womble positively identified as ‘Bubbles.’” State v. Henderson, 937 A.2d 988, 992 (N.J. Super. Ct. App. Div. 2008), aff’d as

modified, 27 A.3d 872 (N.J. 2011).

92 27 A.3d at 880 (internal quotation and citation omitted).

93 Id. All of the photos were headshots of African American men between the ages of

twenty-eight and thirty-five, with short hair, goatees, and, according to Weber, possessing similar facial features. Id.

94 Id.

95 Id.

96 Id.

97 Id.

98 Id.

99 Id.

2020] EYEWITNESS IDENTIFICATION 201

After the New Jersey Supreme Court’s decision in Henderson, on

remand the trial court conducted a hearing, made findings regarding each of

the relevant systems and estimator variables, and then denied the motion to

suppress Womble’s identification. 100

On appeal, that ruling was upheld by

the New Jersey intermediate appellate court. 101

The New Jersey Supreme

Court denied Henderson’s petition seeking further review. 102

Viewed in terms of systems and estimator variables, this is an

extraordinary result. As for estimator variables, when Womble viewed the

perpetrator, he had been drinking and smoking crack, the hallway was dark,

the incident was brief, Womble equivocated about whether he had a good

look at the perpetrator, the lineup did not occur until two weeks after the

shooting, and during the lineup, Womble was initially unable to make an

identification. 103

As for systems variables, Womble was threatened with

prosecution after his initial statement, and Womble could not make an

identification until after he was confronted by the same investigators who

had previously threatened him, in contravention of applicable guidelines

requiring blind lineup administration. 104

If an identification is deemed admissible in the face of so many systems

and estimator variables suggesting unreliability as were present in Henderson

itself, Henderson’s protections may prove illusory. 105

At a minimum, it

seems clear that Henderson erects a highly permeable barrier to the

admission of identifications obtained through suggestive procedures. In

State v. Wright, for example, a New Jersey appellate court upheld a trial

court’s refusal to suppress a concededly suggestive showup identification

made after officers told the victim of an armed robbery that they had arrested

100 See State v. Henderson, 77 A.3d 536, 540–41, 545–56 (N.J. Super. Ct. App. Div.

2013).

101 See id. at 546.

102 State v. Henderson, 91 A.3d 25 (N.J. 2014) (table).

103 See supra text accompanying notes 82–86.

104 See supra text accompanying notes 87–98.

105 E.g., Sandra Guerra Thompson, Daubert Gatekeeping for Eyewitness Identifications,

65 S.M.U. L. REV. 593, 632 (2012) (“At the same time, however, the [New Jersey] cases

exemplify courts’ continuing reluctance to implement reliability gatekeeping as a procedural

norm. Rejecting the recommendations of the Special Master, the New Jersey high court established lofty burdens for defendants to even obtain pretrial hearings, downplayed the need

for defense expert witnesses, and touted jury instructions as a cure-all.”). Notably, the

intermediate appellate court in New Jersey has read Henderson to permit the admission in an

armed robbery case of a showup identification of a handcuffed suspect in police custody by a witness who had also been told that the witness’s stolen cellphone as well as a gun had been

recovered from the vicinity of the suspect. See State v. Wright, 133 A.3d 656, 659–60, 662–

65 (N.J. Super. Ct. App. Div. 2016).

202 ROSENTHAL [Vol. 110

the perpetrator. 106

To be sure, Henderson is sufficiently recent that there are

only a limited number of decisions illustrating its scope, and there are a

handful cases in which a contested identification was suppressed under

Henderson. 107

Still, it is unclear that Henderson has substantially altered the

admissibility of identification testimony in New Jersey. Indeed, as we have

seen, the Henderson standard for excluding an identification ultimately

seems little different from the standard embraced in Manson. 108

The Oregon Supreme Court’s decision in Lawson may have little more

in the way of teeth. Like Henderson, Lawson cautioned that its holding is

unlikely to lead to frequent suppression of evidence. 109

Moreover, in one of

the two cases before the court, it upheld the admission of an identification

even though it involved a showup procedure in which robbery victims viewed

an individual while handcuffed in the back of a police vehicle. 110

Subsequently, the court held that even when a witness identifies an individual

for the first time in what is plainly an unnecessarily suggestive showup

procedure—during her testimony in court—and even though she told police

shortly after the shooting that she did not get a good look at the perpetrators,

the showup identification was nevertheless admissible under Lawson. 111

As for Alaska law, even as it followed Henderson, the Alaska Supreme

Court cautioned, “Although the defendant must only identify a relevant

system variable in order to obtain a hearing, the defendant retains the burden

of proving at that hearing a ‘very substantial likelihood of irreparable

misidentification.’” 112

Similarly, when adopting Henderson, the Connecticut

Supreme Court emphasized that once the state “offer[s] evidence

demonstrating that the identification was reliable,” the burden is on the

defendant to “prove a very substantial likelihood of misidentification,” and

added that the factors courts should consider under Henderson are “generally

106 133 A.3d 656, 662–65 (N.J. Super. Ct. App. Div. 2016); cf. State v. Pressley, 181 A.3d 1017, 1020–21 (N.J. 2018) (upholding admission of undercover officer’s showup

identification occurring shortly after the charged crime).

107 See State v. Drew, No. A-0187-17T4, 2018 WL 2339509 (N.J. Super. Ct. App. Div.

May 24, 2018); State v. Wyles, No. A-3471-16T4, 2017 WL 4558467 (N.J. Super. Ct. App.

Div. Oct. 13, 2017).

108 See supra text accompanying notes 79–81.

109 The court wrote: “[W]e anticipate that the trial courts will continue to admit most

eyewitness identifications . . . . [I]t is doubtful that issues concerning one or more of the

estimator variables that we have identified will, without more, be enough to support an

inference of unreliability sufficient to justify the exclusion of the eyewitness identification.” State v. Lawson, 291 P.3d 673, 697 (Or. 2012).

110 Id. at 681–82, 699–700.

111 See State v. Hickman, 330 P.3d 551, 568–71 (Or. 2014).

112 Young v. State, 374 P.3d 395, 428 (Alaska 2016) (quoting State v. Henderson, 27 A.3d

872, 920 (N.J. 2011) (citation omitted)).

2020] EYEWITNESS IDENTIFICATION 203

comparable” to those employed in federal constitutional law. 113

This

suggests little in the way of meaningful departure from the Manson standard.

Thus, there is considerable doubt as to the practical significance of the

reforms adopted in these jurisdictions.

2. Per se Exclusionary Rules

Per se rules excluding suggestive identifications seemingly offer more

certain protection against the use of suggestive identifications than cases like

Henderson and Lawson. There is, however, reason to doubt that those states

that have purported to adopt per se exclusionary rules have really done so.

For example, even though the Massachusetts Supreme Judicial Court

purported to require the exclusion of all identification evidence obtained

through unnecessarily suggestive procedures, 114

that court later upheld the

admission of identifications obtained by an officer who knew the identity of

the suspect, rather than by a blind administrator—despite acknowledging that

these procedures increase the risk of suggestion 115

—an identification

obtained after a non-blind administrator told the witness that a suspect had

been apprehended before showing the witness a photo array, 116

and a showup

identification of individuals visibly in police custody. 117

All of these seem

like unnecessarily suggestive identifications, and yet none were suppressed.

This is not to say that, in at least some of these cases, the likely effect of the

deviation from practices that minimize the risk of suggestion may have been

small, but it is still hard to square the outcome in these cases with a rule that

purports to suppress all identifications obtained through any form of

unnecessary suggestion.

Similarly, despite its prior holding that purported to brand as

inadmissible all eyewitness identification evidence obtained through

113 State v. Harris, 191 A.3d 119, 131, 146 (Conn. 2018). The court added that inquiry it had adopted was “generally comparable to” the approach taken by federal due process doctrine

and is “merely intended to ‘more precisely define the focus of the relevant inquiry.’” Id. at

136 (internal quotations and citations omitted)).

114 See, e.g., Commonwealth v. Johnson, 650 N.E.2d 1257, 1260 (Mass. 1995) (“The rule of per se exclusion . . . states that the defendant bears the burden of demonstrating, by a

preponderance of the evidence, that the ‘witness was subjected by the State to a confrontation

that was unnecessarily suggestive and thus offensive to due process.’ If this is established,

then the prosecution is barred from introducing that particular confrontation in evidence at trial.” (citations omitted) (quoting Commonwealth v. Botelho, 343 N.E.2d 876, 880 (Mass.

1976)).

115 See Commonwealth v. Silva-Santiago, 906 N.E.2d 299, 311–12 (Mass. 2009).

116 See Commonwealth v. Watson, 915 N.E.2d 1052, 1057–60 (Mass. 2009).

117 See Commonwealth v. Meas, 5 N.E.3d 864, 872–73 (Mass. 2014).

204 ROSENTHAL [Vol. 110

unnecessarily suggestive procedures in People v. Adams, 118

the New York

Court of Appeals subsequently upheld the admission of a showup

identification of a suspect handcuffed in the back of a police car who had

already been identified by another witness, despite acknowledging that this

procedure “[wa]s suggestive and not preferred. It presses judicial tolerance

to its limits.” 119

In another case, the court upheld the use of a showup

procedure in which the suspect was in custody and the identification was

made in the presence of other witnesses, despite acknowledging that “the

better practice when feasible is not to conduct a showup before a group of

witnesses, procedures that are less than ideal may . . . be tolerable in the

interest of prompt identification.” 120

In Massachusetts and New York, in short, it seems that a rule requiring

the suppression of any unnecessarily suggestive identification has merely

meant that those courts will find a great deal of suggestion as necessary. As

for Wisconsin’s supreme court, the court has even more plainly retreated

from a rule that would require the suppression of unnecessarily suggestive

identifications. While seeming to find all unnecessarily suggestive

identifications as inadmissible in its earlier decision in State v. DuBose, 121

the court subsequently held this rule inapplicable to suggestive

identifications made through photographic arrays, 122

suggestive

identifications not arranged by the authorities, 123

and suggestive

identifications that occur during judicial proceedings. 124

In light of these

developments, DuBose seemed ripe for overruling. That is indeed what the

Wisconsin Supreme Court eventually did, finding Dubose unwarranted and

in irreconcilable tension with subsequent Wisconsin precedents. 125

3. Incremental Reforms

As for the rule adopted by the Utah Supreme Court, it is unclear that it

differs from Manson. In terms that seem to track Manson, the court explained

118 423 N.E.2d 379, 383–84 (N.Y. 1981) (“Permitting the prosecutor to

introduce evidence of a suggestive pretrial identification can only increase the risks of convicting the innocent . . . . [I]f the jury finds the in-court identification not entirely

convincing it should not be permitted to resolve its doubts by relying on the fact that the

witness had identified the defendant on a prior occasion if that identification was made under

inherently suggestive circumstances.”).

119 People v. Duuvon, 571 N.E.2d 654, 656–57 (N.Y. 1991).

120 People v. Love, 443 N.E.2d 948, 949 (N.Y. 1982) (memorandum) (citation omitted).

121 699 N.W.2d 582, 591–99 (Wis. 2005).

122 See State v. Drew, 740 N.W.2d 404, 406–09 (Wis. 2007).

123 See State v. Hibl, 714 N.W.2d 194, 199–202 (Wis. 2006).

124 See State v. Ziegler, 816 N.W.2d 238, 256–58 (Wis. 2012).

125 See State v. Roberson, 935 N.W.2d 813, 825–28 (Wis. 2019).

2020] EYEWITNESS IDENTIFICATION 205

that its test “is whether, under the totality of the circumstances, the

identifications were reliable.” 126

The supreme courts of Kansas and Maine,

while expanding the scope of reliability review, have embraced the same

totality-of-the-circumstances test. 127

Moreover, the law in these states has

proven no obstacle to the admission of showup identifications in which a

victim who has witnessed a violent crime is asked to identify a suspect visibly

in police custody. 128

As with Manson, this approach permits the use of

identifications obtained even through unnecessarily suggestive procedures if

a court, with the benefit of hindsight, is willing to deem them reliable.

The North Carolina and Ohio statutes also provide quite limited

protections. In North Carolina, there is no authority that permits suppression

of an identification as long as the trier of fact is made aware of the statutory

violation. 129

Thus far, the Ohio statute has had even less effect; the

intermediate appellate court has interpreted the statute to require suppression

under what is effectively the Manson standard—when “the identification

procedure used was so impermissibly suggestive as to give rise to a very

substantial likelihood of misidentification.” 130

Thus, exclusionary rules that purport to depart from Manson have not

traveled very far. 131

Even these limited reforms, however, have been rejected

by most jurisdictions. Most courts, when invited to depart from Manson as

a matter of state law, have declined to do so. 132

Indeed, Manson’s reliability

126 State v. Guzman, 133 P.3d 363, 367–68 (Utah 2006) (quoting State v. Hubbard, 48

P.3d 953, 963 (Utah 2002) (footnote omitted)).

127 See State v. Corbett, 130 P.3d 1179, 1190–91 (Kan. 2006); State v. Trammell, 92 P.3d

1101, 1107–08 (Kan. 2004); State v. Davis, 191 A.3d 1147, 1155–57 (Me. 2018).

128 See, e.g., State v. Cruz, 307 P.3d 199, 203–04, 208–11 (Kan. 2013) (murder); State v.

Hoffhine, 20 P.3d 265, 266–69 (Utah 2001) (armed robbery); cf. State v. Davis, 191 A.3d 1147, 1155–57 (Me. 2018) (witness had seen booking photo of defendant published in a local

newspaper reporting on defendant’ arrest prior to making identification).

129 See, e.g., State v. Stowes, 727 S.E.2d 351, 357–58 (N.C. Ct. App. 2012); State v.

Howie, No. COA13-553, 2014 WL1047373 at **9–10 (N.C. Ct. App. Mar. 18, 2014).

130 State v. Shaw, 4 N.E.3d 406, 420 (Ohio Ct. App. 2013).

131 For a similar assessment of the limited significance of the departures from Manson undertaken in some states, see Kahn-Fogel, supra note 14, at 160–62.

132 See, e.g., Small v. State, 211 A.3d 236, 244–47 (Md. Ct. App. 2019) (declining to

follow Henderson); Smiley v. State, 111 A.3d 43, 51–52 (Md. 2015) (same); Batiste v. State,

121 So. 3d 808, 855 n.7 (Miss. 2013) (same); People v. Blevins, 886 N.W.2d 456, 462 (Mich. Ct. App. 2016) (same); State v. Moore, No. COA 15-52, 2015 WL 4898121 at *4 (N.C. Ct.

App. Aug. 18, 2015) (same); State v. Discola, 184 A.3d 1177, 1187–89, 1189 n.5 (Vt. 2018)

(declining to follow Massachusetts and New York law).

206 ROSENTHAL [Vol. 110

test for the admission of suggestive eyewitness identification evidence is

utilized by the courts of some forty-one states and the District of Columbia. 133

To what should we attribute the continued vitality of Manson, despite

the many attacks launched against it? Stare decisis is not a satisfactory

answer. As we have seen, Manson’s approach is utilized even by state courts

when applying state law, despite the fact that Manson’s holding on the scope

of the federal constitutional bar on the admission of eyewitness identification

evidence does not prevent state courts from adopting broader rules of

exclusion as a matter of state law. 134

It is to that question that we now turn.

II. THE DIFFICULTIES OF BLACKSTONIAN REFORM

Perhaps the most obvious difficulty a court faces when deciding

whether to adopt a more robust gatekeeping role for eyewitness identification

evidence is identifying a superior alternative to Manson. The costs and

benefits of a different approach to gatekeeping are difficult to assess.

133 See Ex parte Frazier, 729 So. 2d 253, 257 (Ala. 1998); State v. Rojo-Valenzuela, 352

P.3d 917, 919–20 (Ariz. 2015); Kellensworth v. State, 644 S.W.2d 933, 935–36 (Ark. 1983);

People v. Cunningham, 25 P.3d 519, 560–61 (Cal. 2001); Bernal v. People, 44 P.3d 184, 190–

92 (Colo. 2002); Younger v. State, 496 A.2d 546, 550–51 (Del. 1985); Fields v. United States, 484 A.2d 570, 574–75 (D.C.1984); Grant v. Florida, 390 So. 2d 341, 343 (Fla. 1980); Gravitt

v. State, 239 S.E.2d 149, 150–51 (Ga. 1977); State v. Masaniai, 628 P.2d 1018, 1024–26

(Haw. 1981); State v. Buti, 964 P.2d 660, 665–66 (Idaho 1998); People v. Manion, 367 N.E.2d

1313, 1316–17 (Ill. 1977); Slaton v. State, 510 N.E.2d 1343, 1348–49 (Ind. 1987); State v. Neal, 353 N.W.2d 83, 86–87 (Iowa 1984); Moore v. Commonwealth, 569 S.W.2d 150, 153–

54 (Ky. 1978); State v. Brown, 907 So. 2d 1, 16–18 (La. 2005); Webster v. State, 474 A.2d

1305, 1314–16 (Md. 1984); People v. Thomas, 902 N.W.2d 885, 886–87 (Mich. 2017) (order);

State v. Ostrem, 535 N.W.2d 916, 921–22 (Minn. 1995); York v. State, 413 So. 2d 1372, 1382–84 (Miss. 1982); State v. Weaver, 912 S.W.2d 499, 520–21 (Mo. 1996); State v.

Pendergrass, 586 P.2d 691, 695–96 (Mont. 1978); State v. Nolt, 906 N.W.2d 309, 322–23

(Neb. 2018); Gehrke v. State, 613 P.2d 1028, 1029 (Nev. 1980); State v. LaRose, 497 A.2d

1224, 1228–29 (N.H. 1985); Patterson v. LeMaster, 21 P.3d 1032, 1037–39 (N.M. 2001); State v. Harris, 301 S.E.2d 91, 95-96 (N.C. 1983); State v. Juene, No. COA18-526, 2019 WL

189866 at **1–2 (N.C. Ct. App. Jan. 15, 2019); In Re R.W.S., 728 N.W.2d 326, 332–36 (N.D.

2007); Reaves v. State, 649 P.2d 777, 779-80 (Ok. Ct. Crim. App. 1982); Commonwealth v.

Johnson, 139 A.3d 1257, 1278 (Pa. 2016); State v. Austin, 731 A.2d 678, 681–83 (R.I. 1999); State v. Stewart, 272 S.E.2d 628, 629–30 (S.C. 1980); State v. Doap Deng Chuol, 849 N.W.2d

255, 261–62 (S.D. 2014); State v. Ferguson, 741 S.W.2d 125, 126–27 (Tenn. Ct. Crim. App.

1987); Delk v. State, 855 S.W.2d 700, 706–08 (Tex. Ct. Crim. App. 1993); State v. Porter,

103 A.3d 916, 923–25 (Vt. 2014); Delong v. Commonwealth, 362 S.E.2d 669, 674 (Va. 1987); State v. Sanchez, 288 P.3d 351, 378 (Wash. Ct. App. 2012); State v. Kennedy, 249 S.E.2d

188, 188–91 (W. Va. 1978); State v. Roberson, 935 N.W.2d 813, 828 (Wis. 2019); Campbell

v. State, 589 P.2d 358, 362–65 (Wyo. 1979).

134 Cf., e.g., Coleman v. Thompson, 501 U.S. 722, 729 (1991) (“[T]his Court has no power to review a state law determination that is sufficient to support the judgment . . . .”); Minnesota

v. Nat’l Tea Co., 309 U.S. 551, 557 (1940) (“It is fundamental that state courts be left free and

unfettered by us in interpreting their state constitutions.”).

2020] EYEWITNESS IDENTIFICATION 207

A. THE EFFECTS OF PROPHYLACTIC REFORM

Determining whether reforms to eyewitness identification procedures

designed to reduce the risks of suggestion would improve the accuracy of the

criminal process is no easy matter. There is little, if any, reason to believe

that more rigorous identification protocols would improve the accuracy of

eyewitness identification.

1. The Costs and Benefits of More Rigorous Identification Protocols

Critics of eyewitness identification evidence do not claim that it is akin

to the categories of evidence considered unduly likely to unfairly prejudice a

jury, such as evidence of an accused’s supposed propensity to violate the

law. 135

No commentator, for example, has argued for a rule that would bar

all identification evidence as unfairly prejudicial—presumably all agree that

some forms of identification evidence are sufficiently reliable to warrant

admission. Rather, the advocates of more robust judicial gatekeeping take a

surgical approach, arguing that eyewitness identification evidence should be

viewed with special caution when particular factors are present that impinge

on reliability, such as the use of suggestive identification procedures. 136

A meta-analysis of published field and archival studies of lineups

conducted by police in actual cases found that 40.8% of witnesses identified

the suspect, 23.7% of witnesses identified an innocent filler, and 35.5% of

witnesses identified no one. 137

Although there is no way to know whether

the suspects in these lineups were the actual perpetrators, and although most

studies did not track the manner in which the lineup was administered, the

high rate of filler identification suggests that police frequently do not utilize

procedures that minimize the risk of error. As the author of the meta-analysis

observed, “The frequency with which witnesses identified fillers in these

field studies raises the question of whether these eyewitnesses were properly

instructed with the warning that the actual culprit might not be in the lineup

135 See, e.g., Michaelson v. United States, 335 U.S. 469, 475–76 (1948) (“The State may not show defendant’s prior trouble with the law, specific criminal acts, or ill name among his

neighbors . . . . The inquiry is not rejected because character is irrelevant; on the contrary, it

is said to weigh too much with the jury and to so over-persuade them as to prejudge one with

a bad general record and deny him a fair opportunity to defend against a particular charge. ”) (footnotes omitted).

136 See supra text accompanying notes 4–6, 47–51. The evidentiary gatekeeping rules that

have been advocated are canvassed in Part I.B above.

137 Wells, supra note 4, at 268–69, 269 tbl.1.

208 ROSENTHAL [Vol. 110

and whether they understood that they were free to make no

identification.” 138

Experiments under laboratory conditions, in which the simulated

“perpetrator” is known and identification procedures are fully documented,

can offer some insight into the risk of error created when police fail to

minimize the risk of suggestion. In laboratory experiments in which

witnesses viewed simulated events and then are asked to make

identifications, there is ample evidence that error rates rise when precautions

are not taken to prevent suggestion, such as the use of fillers that resemble

the suspect and blind administration. 139

To be sure, there is reason to doubt the reliability of studies conducted

under lab conditions. For example, as we have seen, the stress and fear of

witnessing a violent crime can inhibit a witness’s memory. 140

This is but one

of a plethora of reasons that virtually all scholars who have addressed the

matter have concluded that studies conducted under laboratory conditions

provide limited insight to actual identifications in the field. 141

Indeed, the available data from the field suggest that the costs and

benefits of more rigorous identification protocols thought to minimize the

risk of official suggestion are, at best, unclear. For example, what is perhaps

the leading study on double-blind administration and non-biased witness

instructions in actual lineups using both sequential and simultaneous

identification protocols in four cities, produced the following results:

138 Id. at 269–70. For a discussion of the difficulties in assessing reliability based on the rate at which witnesses identify an individual that the police have designated as the suspect,

see Ruth Horry et al., Archival Analyses of Eyewitness Identification Test Outcomes: What

Can They Tell Us About Eyewitness Memory?, 38 LAW & HUM. BEHAV. 94, 96 (2014); and

Daniel B. Wright et al., Field Studies of Eyewitness Memory, in REFORM OF EYEWITNESS IDENTIFICATION PROCEDURES, supra note 2, at 179, 195–98.

139 See, e.g., Wells, supra note 4, at 265–66 (discussing laboratory research into lineups

in which only the suspect resembles the description of the perpetrator, and the potential for

confirming feedback offered by non-blind lineup administrators).

140 See supra text accompanying note 49. For helpful discussions of the pertinent research,

see IDENTIFYING THE CULPRIT, supra note 4, at 94–96; LOFTUS ET AL., supra note 4, at §§ 2– 9; Kerri L. Pickel, Remembering and Identifying Menacing Perpetrators: Exposure to

Violence and the Weapons Focus Effect, in 2 HANDBOOK OF EYEWITNESS PSYCHOLOGY:

MEMORY FOR PEOPLE 339, 339–47 (Rod C.L. Lindsay et al. eds., 2007); and Ebbe B. Ebbenson

& Vladimir J. Konečni, Eyewitness Memory Research: Probative v. Prejudicial Value, 5 EXPERT EVID. 1, 8–11 (1996).

141 For more elaborate discussions of this problem, see Ebbenson & Konečni, supra note

140, at 4–6; and Wright et al., supra note 138, at 195–98.

2020] EYEWITNESS IDENTIFICATION 209

Table 1142

Administration

Method

Suspect Identified

by Witness

Filler Identified

by Witness

No Identification

by Witness

Simultaneous 26.0% 17.8% 56.2%

Sequential 27.5% 12.3% 60.2%

Prior Archival

and Field Studies

40.8% 23.7% 35.5%

These results reflect a reduction in the rate at which innocent fillers are

identified when compared to prior field and archival studies in which

protocols requiring double-blind administration and non-biased witness

instructions were not employed, but they also reflect even larger reductions

in the rate at which the suspect is identified, as well as a larger increase in

the rate at which witnesses make no identification. 143

The authors argued

that the reduced rate of filler identifications suggests that double-blind

administration reduces the likelihood that an innocent person will be

mistakenly identified. 144

Yet, the rate of filler identifications is a poor proxy

for establishing the rate at which innocent individuals are convicted; after all,

innocent fillers are not likely to be prosecuted since there is, presumably, no

evidence linking them to the crime under investigation. Perhaps even more

important, in this study, the rate at which the suspect was identified dropped

by nearly one-third, and the rate at which no one was identified rose by more

than one third. Because we do not know which of the suspects in the lineups

were actual perpetrators, we cannot know if the reduced rate of

identifications redounded primarily to the benefit of the innocent or the

guilty. Perhaps more rigorous identification protocols reduce the rate of

142 This study reflects the 494 double-blind photo lineups shown to actual witnesses in real cases in four cities (Charlotte-Mecklenburg, North Carolina, Tucson, Arizona, San Diego,

California, and Austin, Texas). The top three rows of Table 1 are derived from the results

reported in Gary L. Wells et al., Double-Blind Photo Lineups Using Actual Eyewitnesses: An

Experimental Test of Sequential Versus Simultaneous Lineup Procedure, 39 LAW & HUM. BEHAV. 1, 7–8, 8 figs. 1 & 2 (2015). For the details of the methodology employed in the study,

see id. at 4–6. The last row reflects the historical rate identified in field studies. See supra text

accompanying note 138.

143 The authors noted the reduced rate at which identifications were made as compared to

prior field and archival studies, and speculated that prior archival studies might be skewed by a failure to document non-identifications in case files, and that the use of double-blind

administration and express instructions to the witness that the culprit might not be present and

that the witness need not make an identification might depress the rate at which identifications

were made. See Wells et al., supra note 142, at 10–11. They also found that the differences in the identification rates between sequential and simultaneous procedures were not statistically

significant. See id. at 7–8.

144 Id. at 12.

210 ROSENTHAL [Vol. 110

mistaken identifications by making it difficult for witnesses to make any

identification at all—even accurate identifications. 145

Also notable is another leading, albeit older, field study. In light of the

tendency of witnesses to make identifications based on a relative judgment

about which face in a lineup most resembles their memory of the

perpetrator, 146

some have advocated the use of sequential lineups, in which

witnesses view faces one at a time, and view all members of a lineup before

they are asked to make an identification—in recognition of the fact that

identification is often based on a relative and not an absolute judgment. 147

A

study utilizing double-blind and sequential photographic lineups in real cases

found that when compared to a then-recent field study used as a baseline,

double-blind and sequential procedures produced a small increase in the rate

that suspects were identified, while the rate at which innocent fillers were

identified declined; but, the rate at which witnesses were unable to make any

identification also rose:

Table 2148

Administration Method Suspect ID Filler ID No ID

Double-blind & Sequential 54% 8% 38%

Archival Baseline 50% 24% 26%

As Table 2 illustrates, in this study, the use of double-blind and

sequential identification procedures did not result in a reduction in the rate of

suspect identifications compared to the archival baseline. The principal

result, instead, was a substitution of filler identifications for no-

identifications. Given that randomly-selected fillers are unlikely to be

145 Cf. Karen L. Amendola & John T. Wixted, The Role of Site Variance in the American

Judicature Field Study Comparing Simultaneous and Sequential Lineups, 33 J. QUANTITATIVE

CRIMINOLOGY 1, 8–12 (2017) (concluding that field study results are attributable to more conservative responding by witnesses in sequential procedures); Rakoff & Loftus, supra note

8, at 94 (“The sequential approach may simply lead to fewer identifications period, reducing

both accurate and inaccurate identifications. At present, the debate and research designed to

inform it continue, suggesting that it is not yet established that one approach is superior to the other.”).

146 See supra text accompanying note 5.

147 See, e.g., R.C.L. Lindsay & Gary L. Wells, Improving Eyewitness Identifications from

Lineups: Simultaneous Versus Sequential Presentation, 70 J. APPLIED PSYCHOL. 556, 562–63

(1985).

148 Table 2 is derived from Amy Klobuchar et al., Improving Eyewitness Identifications: Hennepin County Blind Sequential Lineup Pilot Project, 4 CARDOZO PUB. L., POL’Y & ETHICS

J. 381, 396–98, 398 tbl.2 (2006). For an explication of the methodology of this study, see id.

at 391–95.

2020] EYEWITNESS IDENTIFICATION 211

prosecuted, converting filler identifications into no-identifications is unlikely

to reduce the frequency of false convictions. Procedures that reduce the rate

at which innocent suspects are identified would protect the innocent, but this

study does not enable us to determine whether double-blind and sequential

identifications reduce the rate at which innocent suspects are falsely

identified as perpetrators.

Equally notable was that the study found that when witnesses were

permitted multiple viewings of the photographic lineup, the rate of suspect

identifications increased. 149

Of course, multiple viewings of the lineup

facilitate relative judgments by the witness, as the authors of the study

acknowledged. 150

Perhaps inhibiting relative judgments means that some

witnesses are unable to make any identification—even an accurate one. In

any event, given that the rate at which the suspect was identified did not

significantly change from the archival baseline, these results offer little

reason to believe that double-blind and sequential administration is likely to

reduce the rate of false convictions, which are most likely to result when a

witness identified the suspects, and quite unlikely when the witness identifies

an innocent filler. Perhaps the consistent rate of suspect identification in this

study suggests that double-blind administration makes little difference to

false-conviction rates, which, after all, are not driven by the rate at which

innocent fillers are identified, but instead by the rate at which innocent

suspects are identified. Since this study did not alter the rate at which

suspects are identified, these procedures are unlikely to reduce the rate at

which innocent suspects are wrongly prosecuted and convicted.

Next, consider an Illinois field study involving three cities of varying

populations, which compared identifications made through simultaneous and

sequential, double-blind procedures. 151

149 In particular, when witnesses in the study were permitted to view lineups multiple

times, the rate at which the suspect was identified rose, as did the rate at which innocent fillers

were identified, while the rate at which no identification was made declined, though it

remained higher than the rate of non-identification in the California field study of 26%:

Laps Lineups (n) Suspect ID Filler ID No Choice

1 68 66% 3% 31%

2 42 50% 10% 40%

3 14 50% 14% 36%

4, 5 or 6 4 25% 75%

Id. at 399 tbl.3.

150 Id. at 398.

151 See SHERI H. MECKLENBURG, ILL. ST. POL., REPORT TO THE LEGISLATURE OF THE STATE

OF ILLINOIS: THE ILLINOIS PILOT PROGRAM ON SEQUENTIAL DOUBLE-BLIND IDENTIFICATION

212 ROSENTHAL [Vol. 110

Table 3152

Simultaneous Sequential

n=548 (319) (229)

ID 59.9% 45%

Filler ID 2.8% 9.2%

No ID 37.6% 47.1%

The dramatic reduction in the rate of suspect identifications through

double-blind, sequential procedures again suggests that witnesses have more

difficulty making identifications when they are deprived of the ability to

make relative judgments. To be sure, the Illinois study has been subject to

fierce methodological attack. 153

Still, the evidence that more rigorous

identification protocols reduce the rate at which suspects are identified is

troubling, especially because we cannot know whether the reduced rate of

suspect identifications disproportionately changed to the benefit of the

innocent or the guilty.

Although, as we have seen, studies under laboratory conditions are

problematic, field studies present problems as well. The difficulty with field

studies is that one can never be sure if a witness, by identifying the individual

that the police regard as the suspect, has in fact identified the actual

perpetrator. Laboratory experiments in which researchers know the identity

of a simulated “perpetrator” do not encounter this problem although, as we

have seen, there is reason to be skeptical about eyewitness identifications

under laboratory conditions. 154

PROCEDURES 22–37 (Mar. 17, 2006) http://eyewitness.utep.edu/Documents/IllinoisPilotStudy

OnEyewitnessID.pdf [http://perma.cc/CQA7-DNDC].

152 Id. at 37–38, 38 tbl.3a.

153 See, e.g., James M. Doyle, Learning from Error in American Criminal Justice, 100 J.

CRIM. L. & CRIMINOLOGY 109, 141–43 (2010); Daniel L. Schacter, et al., Studying Eyewitness Investigations in the Field, 32 LAW & HUM. BEHAV. 3 (2008) (arguing study was flawed by

using both double-blind and sequential administration as variables and discussing the

suspiciously low rate of filler identifications in the control group); Nancy K. Steblay, What

We Know: The Evanston Illinois Field Lineups, 35 LAW & HUM. BEHAV. 1 (2011) (discussing evidence that identifications do not reflect random assignment to control and intervention

groups). For defenses of the study, see Sheri H. Mecklenburg et al., The Illinois Field Study:

A Significant Contribution to Understanding Real World Eyewitness Identification Issues, 32

LAW & HUM. BEHAV. 22 (2008); and Stephen J. Ross & Roy S. Malpass, Moving Forward: Response to “Studying Eyewitness Identifications in the Field”, 32 LAW & HUM. BEHAV. 16

(2008).

154 See supra notes 142–143 and accompanying text.

2020] EYEWITNESS IDENTIFICATION 213

A meta-analysis of published laboratory studies of various identification

procedures concluded that the use of lineups, rather than showup

identifications, reduced rates of false identifications while producing

somewhat higher rates at which the perpetrator was identified, while all the

other protocols examined in the literature that are thought to reduce the risk

of suggestion (unbiased instructions to witnesses, sequential lineups, similar-

looking fillers in lineups, and blind administrators), reduced the rate of both

false and correct identifications—usually with larger reductions in the rate of

correct as opposed to false identifications of perpetrators. 155

In particular,

the studies considered in the meta-analysis reflected the following:

Table 4156

Condition Correct

ID Rate

False

ID Rate

Lineup Instructions (n=23)

Biased .59 .15

Unbiased .50 .09

Presentation Format (n=51)

Simultaneous .54 .15

Sequential .43 .09

Lineup Filler Similarity (n=18)

Lower .67 .31

Higher .59 .16

Administrator Influence (n=11)

More .58 .21

Less .45 .11

Showups vs. Lineup (n=15)

Showup .41 .18

Lineup .43 .11

These results, like the studies canvassed above, suggest that more

rigorous identification protocols do not simply reduce error rates, but also

155 See Steven E. Clark, Costs and Benefits of Eyewitness Identification Reform:

Psychological Science and Public Policy, 7 PERSP. PSYCHOL. SCI. 238, 241–44 (2012). To

similar effect, see Clark, supra note 7, at 1119–31; Steven E. Clark et al., Legitimacy,

Procedural Justice, Accuracy, and Eyewitness Identification, 8 U.C. IRVINE L. REV. 41, 67– 77 (2018).

156 This table is derived from Clark, supra note 155, at 242 tbl.2. For a discussion of the

methodology used to identify the studies included in the analysis, see id. at 241, 252–53.

214 ROSENTHAL [Vol. 110

make it more difficult for witnesses to make any identification, whether

accurate or not.

Thus, a tradeoff between reduced rates of false identifications of

innocent suspects and increased rates at which witnesses are unable to make

identifications of guilty perpetrators has been seen both in laboratory and

field studies. Perhaps what some label as suggestive identification

procedures are better characterized as procedures that avoid those

circumstances that make it unduly difficult for the witness to make an

identification. When fillers look much like the suspect, for example, it may

be too difficult for witnesses to select between them.

Consider what may be the simplest case for reform—the use of double-

blind administration to eliminate the risk of administrator bias tainting an

identification. 157

Even in that context, we have no idea what the ratio of false

identifications caused by bias in non-blind administration to the loss of

accurate identifications is. 158

Perhaps blinded procedures create additional

stress for witnesses that inhibits their ability to make an accurate

identification; perhaps what some label as suggestion can also be fairly

characterized as taking care to avoid creating undue stress and difficulty for

witnesses; and perhaps witnesses are far more resistant to being steered

toward identifying an innocent suspect than a guilty one. 159

To this, one might respond that if more rigorous protocols reduce

identification rates, this is only because unreliable identifications are lost. 160

It may also be the case, however, that when identifications protocols become

extremely rigorous, it becomes unduly difficult for some witnesses to make

even reliable identifications. As a committee of the National Research

Council explained, decisions based on memory rest on “two important

parameters: the observer’s memory sensitivity and the degree of evidence

that the observer requires to make an identification.” 161

The committee

elaborated:

157 For a helpful discussion of the case for blind administration, see Margaret Bull Covera

& Andrew J. Evelo, The Case for Double-Blind Administration, 23 PSYCHOL. PUB. POL’Y &

L. 421 (2017).

158 Cf. id. at 424 (referring to “the paucity of studies available”).

159 Cf. Clark et al., supra note 155, at 73–77 (suggesting that non-blind administrators may be more sensitive to witness reactions and accordingly more helpful to witnesses while rarely

able to steer witnesses toward identifying innocent suspects).

160 See, e.g., Risinger, supra note 18, at 343–47 (arguing that identifications obtained by

non-blind administrators are tainted by unnecessary suggestion and therefore lack probative

value); Gary L. Wells et al., Eyewitness Identification Reforms: Are Suggestiveness-Induced Hits and Guesses True Hits?, 7 PERSP. PSYCHOL. SCI. 264, 265–66 (2012) (arguing that

identifications produced by suggestive procedures should not be regarded as legitimate).

161 IDENTIFYING THE CULPRIT, supra note 4, at 80 (parentheticals omitted).

2020] EYEWITNESS IDENTIFICATION 215

If a witness sets a high bar for acceptable evidence—a conservative bias—then he or

she will be unlikely to select anyone from the lineup (low pick frequency), meaning

that they will have more misses (will be more likely to fail to select the suspect because

they are less likely to make a selection at all) and fewer false alarms.

Conversely, if a witness sets a low bar for acceptable evidence—a liberal bias—then

she or he will be more likely to select from the lineup (a high pick frequency), meaning

that he or she will have more hits and will make more false identifications.162

Thus, witnesses whose internal threshold for making an identification is

relatively low would likely produce a higher error rate regardless of the

identification protocols employed. We have little idea, however, how to

identify the witnesses that utilize low internal thresholds for making

identifications and therefore present an elevated risk of error. Conversely,

witnesses who utilize a relatively high internal threshold might find more

rigorous identification protocols unduly daunting. Given these complexities,

outside of pristine laboratory conditions we have no way of knowing at what

rate accurate and false identifications are lost when more rigorous

identification protocols are employed.

Perhaps more important, even an identification based on a witness’s

relative judgment that a given suspect resembles the witness’s recollection of

the perpetrator might, coupled with other evidence, amount to reliable proof

of guilt. 163

There is evidence from field studies, for example, indicating that

identifications produced by simultaneous procedures—despite their greater

likelihood to be based on relative judgments—were associated with stronger

independent evidence of the suspect’s guilt than identifications made using

sequential procedures. 164

Beyond that, if in the real world the perpetrator is likely to be present in

lineups, then an identification based on a witness’s relative judgment about

which face most resembled the perpetrator could have considerable probative

value, especially when combined with independent evidence of guilt. 165

In

162 Id. at 82.

163 For an effort to demonstrate, using simulations, that eyewitness identifications based on relative judgments may not be materially less accurate than identifications based on

absolute judgments, see Steven E. Clark et al., Probative Value of Absolute and Relative

Judgments in Eyewitness Identification, 35 LAW & HUM. BEHAV. 364 (2011).

164 See Amendola & Wixted, supra note 145, at 12–18 (analysis of field data from

Charlotte-Mecklenburg, Tucson, Arizona, San Diego, and Austin suggesting that simultaneous identifications occurred in cases with stronger independent evidence of guilt and

higher likelihood of an adjudication of guilt); John T. Wixted, et al., Estimating the Reliability

of Eyewitness Identifications from Police Lineups, 113 PROC. NAT’L ACAD. SCI. 304, 308–09

(2016) (field study in Houston finding stronger independent corroborating evidence of guilty for suspects identified through simultaneous rather than sequential procedures).

165 Cf. Michael A. Palmer & Neil Brewer, Sequential Lineup Presentation Promotes Less-

Biased Criterion-Setting But Does Not Improve Discriminability, 36 LAW & HUM. BEHAV.

216 ROSENTHAL [Vol. 110

other cases, conversely, highly suggestive identification procedures could

inject an unacceptable risk of error if a prosecution is based on little more

than a superficial resemblance between a suspect and a perpetrator. 166

Accordingly, assessing the costs and benefits of procedures that

facilitate identifications, even if they also involve a potential for suggestion,

most likely requires knowing the frequency at which identifications are made

when (1) the actual perpetrator is in the lineup, and (2) there is independent

evidence of guilt that, when combined with the identification, yields a

reliable case against the suspect. In the real world, however, we have no idea

how often that occurs. 167

In sum, the available data suggests that more rigorous identification

protocols involve some kind of rough tradeoff between reduced rates in

which innocent suspects are falsely identified and increased rates at which

witnesses are unable to identify the guilty. Beyond that, the data are noisy,

sometimes inconsistent, and provide nothing approaching a clear indication

that reforms that reduce the risk of suggestion are likely to have a meaningful

effect on the rate of false identifications—much less benefits that exceed their

costs. The data are chaotic, and the state of our knowledge about eyewitness

identification reform remains primitive. Perhaps, over more time than is

reflected in the studies canvassed above, police would learn to administer

more rigorous identification protocols in a way that would reduce their costs;

but, at present, there is no reliable evidence to support such speculation. 168

247, 253–54 (2012) (“[I]n situations where culprit-present lineups are far more common than culprit-absent lineups, accuracy is maximized when responding is lenient, rather than

unbiased. Conversely, when the base rate of culprit-present lineups is low, accuracy is

maximized by conservative responding. Because the base rate of culprit-present lineups in

actual police investigations is not known, we cannot be certain whether less-biased responding will produce greater accuracy in these settings.”).

166 Cf. Risinger, supra note 18, at 354 (“[A]t least in the case of targets who have been

selected only because of a claimed resemblance to the perpetrator, the dangers of relative

judgment are so great that simultaneous presentation ought never to be undertaken.”).

167 Cf. id. at 356 (“Unlike the situation where a target is chosen to be put in a lineup merely

on the basis of some judgment of physical resemblance to the perpetrator, the normal lineup situation involves putting the target into the lineup for other reasons bearing on a likelihood

of guilt greater than a random draw from a large population, such as a tip, etc. How often does

such independent evidence result in the true perpetrator being in the lineup, and how often

not? The answer is, we really don’t know. There simply is no good empirical evidence on the issue.”) (footnotes omitted).

168 In this connection, it is worth noting that even after police have had decades to adjust

to the rule requiring that they advise individuals of their rights during custodial interrogation

announced in Miranda v. Arizona, 384 U.S. 436 (1966), there is evidence that Miranda continues to reduce confession and clearance rates in criminal investigations. See Paul G.

Cassell & Richard Fowles, Still Handcuffing the Cops? A Review of Fifty Years of Empirical

Evidence of Miranda’s Harmful Effects on Law Enforcement, 97 B.U. L. REV. 685 (2017).

2020] EYEWITNESS IDENTIFICATION 217

Perhaps one day we will have carefully controlled studies demonstrating

the effect of more rigorous identification protocols on the rates of both

accurate and false identification. Yet, even if we could be confident about

the rate at which more rigorous identification protocols screen out false rather

than accurate identifications, the question remains: “What should the

exchange rate be for correct identifications lost versus false identifications

avoided?” 169

Even this formulation, however, likely understates the problem.

Given the difficulties in quantifying the costs of both wrongful acquittals and

convictions, cost-benefit analysis in this arena presents formidable

difficulties. 170

2. The Problematic Case for Blackstonian Prophylactic Rules

If jurors could accurately assess the reliability of an eyewitness

identification, the use of potentially suggestive procedures would not be a

problem. We would be able to avoid the potential loss of accurate

identifications from rigorous identification protocols by permitting

potentially suggestive procedures, confident in the jury’s ability to assess the

risk of error created by those procedures. As we have seen, however, the

available research suggests that juries overestimate the reliability of

eyewitness identifications once admitted in evidence. 171

Although more extensive use of expert testimony and cautionary jury

instructions might ameliorate this problem, the available research on mock

juries suggests that providing them with additional information about the

perils of eyewitness identification has limited effects. 172

If jurors’ intuitions

lead them to place great weight on the testimony of an eyewitness with no

obvious motive to lie, it is far from clear that a counterintuitive lecture on the

psychology of eyewitness identification is likely to eliminate the problem. 173

169 Clark, supra note 155, at 248.

170 Cf. IDENTIFYING THE CULPRIT, supra note 4, at 77–86, 104, 117–18 (the committee

concluded that in light of the difficulties in assessing the costs and benefits of sequential identification protocols compared to sequential protocols, it could not make a recommendation

between the two.).

171 See supra text accompanying notes 7–8.

172 See, e.g., CUTLER & PENROD, supra note 6, at 257–63 (summarizing research); Kahn-

Fogel, supra note 14, at 119–20 (same); Rakoff & Loftus, supra note 8, at 96–97; see also

Wells & Quinlivan, supra note 6, at 21 (“Whether jury instructions . . . will have much impact on the jury is an open question, but it is likely to serve a deterrent function because prosecutors,

who are motivated to keep such instructions away from the jury, will likely help bring pressure

back on their police departments to avoid suggestive procedures in the future.”).

173 Cf. Peter J. Cohen, How Shall They Be Known? Daubert v. Merrell Dow Pharmaceuticals and Eyewitness Identification, 16 PACE L. REV. 237, 272–73 (1996) (“There

is no scientific evidence that cautionary jury instructions, given at the end of what might be a

long and fatiguing trial, and buried in an overall charge by the court, are effective. A powerful

218 ROSENTHAL [Vol. 110

And even if jurors heed the instructions, that may create new problems;

studies of the jury instructions utilized in New Jersey since the Henderson

decision indicate that the new instructions cause mock jurors to become more

skeptical of all eyewitness identifications, regardless of the strength of the

evidence. 174

Because the risk of error in jurors’ assessments of eyewitness

identification may be ineradicable, a more robust rule of exclusion triggered

by a failure to observe prophylactic safeguards against potentially suggestive

identification procedures might seem the only effective way to reduce the

rate of wrongful convictions attributable to eyewitness identifications. 175

Unlike Manson’s exclusionary rule, such an approach is not premised on a

finding that a particular identification is unreliable and therefore likely to

eyewitness’ testimony may be so firmly embedded in the jurors’ minds that the court’s

instructions days or weeks later may be unable to undo potential prejudice.”).

174 See Rakoff & Loftus, supra note 8, at 96–97 (concluding that the studies of the New Jersey instructions “suggest that such instructions may not adequately serve their intended

function of enabling jurors to discriminate more accurately between reliable and unreliable

eyewitness testimony”).

175 Likely the best-known instance of what could be regarded as a prophylactic rule of constitutional law is the holding in Miranda v. Arizona, 384 U.S. 436 (1966), rendering

statements made during custodial interrogation inadmissible absent the provision of specified

warnings and a valid waiver of the Fifth Amendment right to be free from compelled self-

incrimination and its progeny. The Court and its members have sometimes characterized the rules derived from Miranda as prophylactic in character. See, e.g., Maryland v. Shatzer, 559

U.S. 98, 103–06 (2010); United States v. Patane, 542 U.S. 630, 638–41 (2004) (plurality

opinion); Chavez v. Martinez, 538 U.S. 760, 770–73 (2003) (opinion of Thomas, J.); Davis v.

United States, 512 U.S. 452, 457–58 (1994); Oregon v. Elstad, 470 U.S. 298, 306–08 (1985); New York v. Quarles, 467 U.S. 649, 654–58 (1984); Michigan v. Tucker, 417 U.S. 433, 438–

46 (1974); cf. Dickerson v. United States, 530 U.S. 428, 442 (2000) (“In Miranda, the Court

noted that reliance on the traditional totality-of-the-circumstances test raised a risk of

overlooking an involuntary custodial confession, a risk that the Court found unacceptably great when the confession is offered in the case in chief to prove guilt. The Court therefore

concluded that something more than the totality test was necessary.”) (citations omitted).

Many commentators have defended Miranda in terms of prophylaxis. See, e.g., Evan H.

Caminker, Miranda and Some Puzzles of “Prophylactic” Rules, 70 U. CIN. L. REV. 1, 9–20 (2001); Yale Kamisar, Confessions, Search and Seizure and the Rehnquist Court, 34 TULSA

L.J. 465, 471–76 (1999); Klein, supra note 73, at 480–88; David A. Strauss, The Ubiquity of

Prophylactic Rules, 55 U. CHI. L. REV. 190, 190–95, 208–09 (1988); Stephen J. Schulhofer,

Reconsidering Miranda, 54 U. CHI. L. REV. 435, 447–53 (1987). Others, however, deny the legitimacy of Miranda as a prophylactic rule of constitutional law. See, e.g., JOSEPH D. GRANO,

CONFESSIONS, TRUTH, AND THE LAW 173–98 (1996). Still others doubt that Miranda is

properly characterized as prophylactic in character, rather than as articulating a judicially

administrable rule for identifying the presence of compelled self-incrimination within the meaning of the Fifth Amendment. See, e.g., Mitchell N. Berman, Constitutional Decision

Rules, 90 VA. L. REV. 1, 114–36 (2004); Lawrence Rosenthal, Compulsion, 19 U. PA. J.

CONST. L. 889, 941–53 (2017).

2020] EYEWITNESS IDENTIFICATION 219

produce a false conviction, but instead on reducing what is thought to be an

elevated risk of error that exists when suggestive procedures are used to

obtain the identification. Commentators advocate the use of prophylactic

rules when the error rate in case-by-case adjudication is high and the benefits

of prophylaxis are therefore likely to exceed its costs. 176

As we have seen, however, there is reason to believe that requiring the

use of more rigorous identification protocols to minimize the risk of

suggestion is likely to increase the rate at which accurate identifications are

lost, perhaps to an even greater extent than the rate at which false convictions

are reduced. Accordingly, a more robust, prophylactic exclusionary rule

might not decrease the overall error rate, although it could well reallocate

error by increasing the rate at which witnesses fail to identify the actual

perpetrator as it decreases the rate of false identifications. If prophylaxis

does not reduce the overall error rate in assessing identification evidence,

however, one could question the justification for a prophylactic rule requiring

the exclusion of identification evidence because such evidence is thought to

present an elevated risk of error. The available empirical evidence canvassed

in Part II.A.1 above does not permit any confident conclusion that a

prophylactic rule would reduce the overall error rate in assessing the

reliability of eyewitness identifications when compared to Manson’s totality-

of-the-circumstances test.

A Blackstonian response to these uncertainties is that the law—perhaps

even the Constitution—tells us how to allocate the risk of error. The

Supreme Court, for example, justified its holding that due process of law

requires that the prosecution prove the defendant’s guilt beyond a reasonable

doubt by reference to the deeply rooted aversion to conviction of the

innocent. 177

And as we know, Blackstone’s ratio argues against the

admission of evidence likely to produce wrongful convictions because “it is

better that ten guilty persons escape, than that one innocent suffer.” 178

As

one commentator put it:

[W]e embrace the value preference expressed by Blackstone’s ratio . . . . While that

ratio is not meant to create a rigid mathematical formula—indeed the acceptable ratio

of wrongful convictions to failures to convict cannot be set with any mathematical

176 For defenses of prophylactic rules along these lines, see, for example, Caminker, supra

note 175, at 6–9; Brian K. Landsberg, Safeguarding Constitutional Rights: The Uses and

Limits of Prophylactic Rules, 66 TENN. L. REV. 925, 949–64 (1999); Kermit Roosevelt III, Constitutional Calcification: How the Law Becomes What the Court Does, 91 VA. L. REV.

1649, 1668–72 (2005); and Strauss, supra note 175, at 195–208.

177 See In re Winship, 397 U.S. 358, 361–64 (1970).

178 4 BLACKSTONE, supra note 17, at *352.

220 ROSENTHAL [Vol. 110

precision—the maxim does at least express a value preference that has been

incorporated into constitutional doctrine.179

Blackstone’s ratio is not uncontroversial; it is far from clear that it

accurately captures the costs and benefits of error in the criminal justice

system. 180

Even among those who favor some type of ratio to protect the

innocent, there is no agreement on whether the correct ratio is 10-1 or

something else. 181

One commentator, for example, offered a “Reform Ratio”

that endeavors to acknowledge the costs of failing to convict the guilty:

Any wrongful conviction that can be corrected or avoided without allowing more than

one or two perpetrators of similar crimes to escape, ought to be corrected or avoided;

in addition, system alterations (reforms, if you will) that there is good reason to believe

will accomplish this ought to be embraced. 182

Given the limited available data on the efficacy of eyewitness

identification reforms canvassed above, however, it is doubtful that any of

them satisfy any plausible reform ratio. In any event, even for those

sympathetic to the Blackstonian preference from wrongful acquittal over

wrongful convictions, invoking the Blackstonian ratio to support the case for

a more robust, prophylatic eyewitness-identification exclusionary rule would

considerably oversimplify matters.

A systemic protection for the innocent is already built into the heavy

burden of proof that the prosecution must shoulder in a criminal case; that

burden itself reflects a preference for false acquittals over wrongful

convictions. 183

Whether additional protections are required when it comes to

eyewitness identification evidence is, however, a separate question. After

all, neither the burden of proof nor the general acceptance of the desire to

minimize the rate of false convictions has produced the view that every

conceivable precaution must be taken to prevent the conviction of the

179 Keith A. Findley, Toward a New Paradigm of Criminal Justice: How the Innocence

Movement Merges Crime Control and Due Process, 41 TEX. TECH L. REV. 133, 136 (2008)

(footnotes omitted).

180 For critiques of the Blackstone ratio arguing that it understates the costs of a ratio

skewed in favor of acquittal, see, for example, Ronald J. Allen & Larry Laudan, Deadly

Dilemmas, 41 TEX. TECH L. REV. 65 (2008); and Daniel Epps, The Consequences of Error in

Criminal Justice, 128 HARV. L. REV. 1065 (2015). For a rebuttal, see Marvin Zalman, The Anti-Blackstonians, 48 SETON HALL L. REV. 1319 (2018).

181 For a survey of the diverse views on this point, see Alexander Volokh, N Guilty Men,

146 U. PA. L. REV. 173 (1997).

182 Risinger, supra note 18, at 360.

183 See text accompanying note 19.

2020] EYEWITNESS IDENTIFICATION 221

innocent. 184

Nor has it taken the view that evidence may not be introduced

if there is some risk that it is unreliable. 185

Indeed, we have little idea what the error rate is for most types of

evidence. Most evidence is not infallible—in fraud cases, sometimes alleged

victims mischaracterize what they were told, innocently or intentionally; in

arson cases, sometimes circumstantial evidence of the defendant’s financial

distress will not always accurately indicate that the defendant set a fire to

collect insurance; in rape cases, sometimes alleged victims lie or are

mistaken. The fact that a type of evidence of guilt is not infallible is no reason

for its exclusion. After all, it is a perilous enterprise to assess in isolation the

reliability of any evidence. Many types of evidence are of dubious reliability

in the abstract, but when combined with the other evidence in the case, might

produce what a jury could justifiably regard as a compelling case.

Accordingly, it is difficult to assess the reliability of any type of

evidence in isolation. A fraud victim’s testimony that she was swindled by

a wealthy and respected investment advisor might, standing alone, seem

unreliable in light of the victim’s financial motive to gain a restitution

payment; but in the face of compelling corroboration, the very same

testimony might come to be seen as credible. The law does not insist that

each individual piece of evidence offered against a defendant constitute

reliable proof of guilt; the reliability of any particular item of evidence is,

instead, appropriately assessed in light of the totality of the proof. 186

184 See, e.g., Patterson v. New York, 432 U.S. 197, 208 (1977) (“While it is clear that our

society has willingly chosen to bear a substantial burden in order to protect the innocent, it is equally clear that the risk it must bear is not without limits . . . . Due process does not require

that every conceivable step be taken, at whatever cost, to eliminate the possibility of

convicting an innocent person.”).

185 Cf. Larry Laudan, Eyewitness Identifications: One More Lesson on the Costs of

Excluding Relevant Evidence, 7 PERSP. PSYCHOL. SCI. 272, 272 (2012) (“[T]here is no rule of admissibility that will not occasionally lead to the conviction of an innocent defendant.

Confessions are sometimes false. Should we exclude them? Expert testimony is occasionally

mistaken. Should it be excluded? . . . . [T]he quest for evidence that infallibly indicates guilt

(or innocence) is a snark hunt. It is provable in principle that there is no rule of evidence or procedure that will not occasionally lead to a false conviction (or a false acquittal). The fact

that relevant evidence leads to fallible inferences is no argument for the former’s exclusion.”).

186 See, e.g., Bourjaily v. United States, 483 U.S. 171, 179–80 (1987) (“[I]ndividual pieces

of evidence, insufficient in themselves to prove a point, may in cumulation prove it. The sum of an evidentiary presentation may well be greater than its constituent parts. Taken together,

these two propositions demonstrate that a piece of evidence, unreliable in isolation, may

become quite probative when corroborated by other evidence.”); cf. Victor v. Nebraska, 511

U.S. 1, 16 (1994) (upholding a jury instruction defining proof beyond “reasonable doubt” as “that state of the case which, after the entire comparison and consideration of all the

evidence, leaves the minds of the jurors in that condition that they cannot say they feel an

abiding conviction, to a moral certainty, of the truth of the charge.”) (emphasis in original).

222 ROSENTHAL [Vol. 110

The likelihood that prophylactic rules that abjure inquiry into the totality

of the circumstances will result in the exclusion of highly probative evidence,

even when obtained through potentially suggestive procedures, may well

explain difficulties courts have had in applying exclusionary rules purporting

to be more robust than Manson. Consider Henderson. Recall that prior to

Womble’s contested identification of Henderson, Womble identified the

accomplice of the man he later identified as Henderson—Clark—and Clark

then identified Henderson as the individual who accompanied him to the

scene of the crime. 187

Subsequently, upon his arrest, Henderson “admitted to

the police that he had accompanied Clark to the apartment where Harper was

killed, and heard a gunshot while waiting in the hallway. But [Henderson]

denied witnessing or participating in the shooting.” 188

Accordingly, there

was compelling corroboration for Womble’s contested identification of

Henderson, including Henderson’s own admission.

As we have seen, it is difficult to explain the ultimate decision to permit

the use of Womble’s contested identification in Henderson in terms of the

systems and estimator variables that were supposed to be the basis for

applying the prophylactic rule fashioned by the New Jersey Supreme

Court. 189

The decision to admit Womble’s identification is far more easily

explained by the ample corroborative evidence demonstrating that Womble’s

identification of Henderson was reliable—not only did Clark (the shooter)

also place Henderson at the apartment, but Henderson admitted he was there.

It is difficult to conclude, in light of this evidence, that Womble’s

To be sure, in the context of the Sixth Amendment right of an accused “to be confronted with

the witnesses against him,” U.S. CONST. amend. VI, the Supreme Court has rejected

consideration of corroborative evidence when assessing the admissibility of statements made

in the absence of confrontation. See, e.g., Idaho v. Wright, 497 U.S. 805, 823 (1990) (“[T]he use of corroborating evidence to support a hearsay statement’s ‘particularized guarantees of

trustworthiness’ would permit admission of a presumptively unreliable statement by

bootstrapping on the trustworthiness of other evidence at trial, a result we think at odds with

the requirement that hearsay evidence admitted under the Confrontation Clause be so trustworthy that cross-examination of the declarant would be of marginal utility.”).

Confrontation Clause jurisprudence, however, presents a quite different problem from the

admissibility of identification evidence when the witness making the identification is subject

to cross-examination; as the Court later explained precisely because the Confrontation Clause identifies confrontation as a precondition for the admission of evidence: “Where testimonial

statements are at issue, the only indicium of reliability sufficient to satisfy constitutional

demands is the one the Constitution actually prescribes: confrontation.” Crawford v.

Washington, 541 U.S. 36, 68–69 (2004). If the confrontation requirement is satisfied, however, the Constitution prescribes no other test for assessing the reliability of evidence

beyond the inquiry into the totality of the circumstances described above.

187 See supra text accompanying notes 87–90.

188 State v. Henderson, 27 A.3d 872, 880 (N.J. 2011).

189 See supra text accompanying notes 56–62.

2020] EYEWITNESS IDENTIFICATION 223

identification was unreliable or somehow otherwise likely to lead the jury

into error, even if the identification was obtained through a potentially

suggestive procedure in which a blind-administration protocol was

compromised.

For another example, consider Commonwealth v. Meas. 190

Despite

purporting to require the exclusion of all identifications obtained through

unnecessarily suggestive procedures, 191

in Meas the Massachusetts Supreme

Judicial Court upheld the use of showup identifications of a suspect visibly

in police custody. In this case, the corroboration included evidence that the

suspect had been stopped in a vehicle matching the description and

displaying a similar license plate number to that which had been provided to

the police by witnesses, shortly after and in the vicinity of a shooting; when

stopped, a loaded gun was found on the vehicle’s floor and a spent shell

casing was found where the suspect had been seated; the spent casing in the

vehicle and another found at the scene of the crime were traced to the firearm

found in the vehicle; and another of the vehicle’s occupants, after pleading

guilty to being an accomplice after the fact, testified that he witnessed the

defendant shoot the victim. 192

There may have been no good reason that the

police failed to arrange a lineup rather than using a suggestive showup

procedure in Meas, but given the strength of the evidence, the identification

created scant risk of error.

Similarly, despite its earlier holding purporting to exclude all

unnecessarily suggestive identifications, 193

the New York Court of Appeals

upheld a showup identification of an individual visibly in police custody after

he had been apprehended near the scene of a robbery of a dry cleaner,

corroborated by the independent identification of the manager who had

chased the robber and caught up with police as they apprehended the suspect,

and corroborated as well by the defendant’s subsequent confession. 194

In each of these cases, adherence to a prophylactic rule requiring the

exclusion of any identification obtained through unnecessarily suggestive

procedures would have been more likely to detract from, rather than enhance,

the reliability of the fact-finding process. Once the contested identification

is considered in light of the other evidence in the case, there is little reason

to doubt its reliability. The rigidity of a prophylactic rule that would exclude

the identification without reference to the other evidence in the case that

bears on the perpetrator’s identity may instead increase the risk of error.

190 5 N.E.3d 864 (Mass. 2014).

191 See Commonwealth v. Johnson, 650 N.E.2d 1257, 1259–65 (Mass. 1995).

192 Meas, 5 N.E.3d at 869–75, 880–81.

193 See People v. Adams, 423 N.E.2d 379, 383–84 (N.Y. 1981).

194 See People v. Duuvon, 571 N.E.2d 654, 655–56 (N.Y. 1991).

224 ROSENTHAL [Vol. 110

In any event, even when prophylactic safeguards are taken, the

admission of eyewitness identification evidence still creates a risk of

wrongful conviction; in the studies canvassed above of identification

protocols designed to reduce the risk of suggestion, none of the protocols was

able to drive the rate of false identifications to zero or thereabouts. 195

Surely

that does not suggest that all identification evidence must be barred because

of the risk of false convictions that it induces—at some unknown ratio to the

increased rate of false acquittals—into the criminal process.

To be sure, sometimes it is apparent that identification evidence is

highly unreliable and should be excluded for that reason. Perhaps Foster is

such a case; investigators undertook something of a campaign of suggestion

directed at a witness who had repeatedly failed to identify Foster, and there

seems to have been little evidence pointing to Foster beyond the contested

identification. 196

Still, in all but the clearest cases—those where there is little,

if any, evidence of identity beyond a highly suggestive identification—we do

not know the extent to which the admission of any particular evidence

increases the likelihood of a wrongful conviction. In most cases, moreover,

there is no ready vehicle for assessing the risk of error injected by eyewitness

identification evidence, or any other type of evidence. Most evidence is not

infallible; surely it is rare that the admission of any type of evidence of guilt

creates no risk of a wrongful conviction. There is, accordingly, no logical

stopping point for exclusion of evidence that gives rise to a risk of wrongful

convictions; this risk inheres in virtually all evidence.

We could reduce the rate of wrongful convictions to something

approaching zero if we required the prosecution to use only the most

unassailable types of evidence, impeccably corroborated—perhaps only

cases involving videotapes of the offense and an ensuing, independent, and

concededly voluntary confession—but, even most Blackstonians would

likely view the resulting reduction in the rate at which the prosecution can

convict the guilty as unacceptable. Surely reforms that increase the rate of

false acquittals are justified only if they produce a sufficiently large reduction

in the rate of false convictions to justify the resulting tradeoff. 197

195 See supra Part II.A.

196 See supra text accompanying notes 35–37.

197 Cf. Paul G. Cassell, Overstating America’s Wrongful Conviction Rate? Reassessing

the Conventional Wisdom About the Prevalence of Wrongful Convictions, 60 ARIZ. L. REV.

815, 855 (2018) (“[M]ore rigid requirements for a valid eyewitness identification may protect some innocent people from being mistakenly identified, but at the cost of preventing some

guilty people from being properly identified. In assessing the tradeoffs involved in such

reforms, the size of relative risks does matter.”) (footnote omitted).

2020] EYEWITNESS IDENTIFICATION 225

Indeed, the Blackstone ratio itself implies an awareness of a tradeoff—

it may not be possible to minimize the rate of false convictions without

unacceptably increasing false acquittals. Thus, even for Blackstonians,

simply minimizing the rate of false convictions will not do. A fully informed

assessment of a prophylactic rule requires consideration of not only the

reduction in the rate of false convictions, but the potential reduction of the

rate at which the guilty are convicted. 198

And, given the limited state of our

knowledge about the ratio between false acquittals and convictions produced

by pretty much all types of evidence, the best Blackstonians can do is insist

on the stringent burden of proof in criminal cases. Assessing the error rate

in each type of evidence offered by the prosecution to determine whether

prophylaxis is justified is an impossible task.

Even a narrower claim that eyewitness identification evidence presents

particular risks that warrant a particularly stringent rule of exclusion remains

problematic. As we have seen, inaccurate eyewitness identifications have

been identified as a leading cause of false convictions. 199

This point,

however, is not as compelling as it might at first blush seem.

Although we can calculate the number of times that a defendant

convicted on the basis of eyewitness identification evidence was later

exonerated, we do not have reliable data about the number of accurate

convictions based on eyewitness identifications; as a result, we cannot know

that the error rate in these cases is unusually high. 200

Moreover, exonerations

tend to cluster in the types of cases where DNA evidence can conclusively

establish the identity of the perpetrator, such as cases resting on eyewitness

identification evidence. 201

The rate of false convictions may be as high or

higher in cases resting on accomplice testimony, the uncorroborated

testimony of a victim, or circumstantial evidence; it is, however, quite

difficult to ascertain the actual rate of false conviction in these cases. One

198 Cf. Manson v. Brathwaite, 432 U.S. 98, 112–13 (1977) (“[T]he per se approach

[requiring the exclusion of all evidence obtained by unnecessarily suggestive procedures] suffers serious drawbacks. Since it denies the trier reliable evidence, it may result, on occasion,

in the guilty going free. Also, because of its rigidity, the per se approach may make error by

the trial judge more likely than the totality approach. And in those cases in which the

admission of identification evidence is error under the per se approach but not under the totality approach, cases in which the identification is reliable despite an unnecessarily

suggestive identification procedure reversal is a Draconian sanction.”) (footnote and citations

omitted).

199 See supra text accompanying note 2.

200 See, e.g., Jennifer E. Laurin, Still Convicting the Innocent, 90 TEX. L. REV. 1473, 1489

(2012) (reviewing GARRETT, supra note 15) (making this point).

201 See, e.g., Gross, supra note 2, at 766 (“If, somehow, DNA permitted us to identify

robbers as effectively as it identifies rapists, we might have over 800 robbery exonerations

rather than 100.”).

226 ROSENTHAL [Vol. 110

survey, for example, made a powerful case that given the fallibility in

witnesses’ memories for conversations and jurors’ tendency to believe

confident witnesses, cases resting on the conversational memories of

witnesses endeavoring to recall statements attributed to the defendant likely

produce more false convictions than cases resting on eyewitness

identifications. 202

Given the advent of DNA evidence, false convictions are more readily

detectable in eyewitness identification cases than many others, but that does

not mean that we can reliably conclude that eyewitness identification

evidence represents anything like a unique problem of reliability. Indeed,

since the advent of DNA evidence capable of identifying false

identifications, perhaps the rate of wrongful convictions in such cases is

likely to be lower in the future than in other cases in which DNA evidence

has less utility, such as those involving conversational memory. 203

Accordingly, although courts may have a limited ability to identify

unreliable identifications under Manson’s totality of the circumstances

approach, it is unclear that there is any prophylactic alternative likely to

improve matters. A prophylactic rule requiring the use of rigorous

identification protocols does not even attempt to identify unreliable

identifications, but instead identifies what is regarded as a proxy for

reliability—the procedures used to obtain identifications. But because a

prophylactic rule does not examine the other evidence in the case that might

corroborate an identification, it may produce a higher error rate than Manson.

Moreover, to justify prophylaxis on the Blackstonian ground that every

precaution should be undertaken to exclude evidence that gives rise to a risk

of false conviction is to adopt a rationale with no logical stopping point—

one that could be used to bar most types of evidence.

202 See Steven B. Duke et al., A Picture’s Worth a Thousand Words: Conversational Versus Eyewitness Testimony in Criminal Convictions, 44 AM. CRIM. L. REV. 1, 6–45 (2007).

There have been a number of efforts to 9estimate an overall rate of false conviction; see, e.g.,

SANGERO, supra note 4, at 8–14 (estimating the rate of false convictions at 5–10%); Allen &

Laudan, supra note 180, at 68–71 (estimating an overall wrongful conviction rate of 0.84%); Cassell, supra note 197, at 846–48 (estimating an overall wrongful conviction rate of 0.016–

0.062%); Gross, supra note 2, at 784–85 (estimating an error rate for death sentences of 4.1%

and for other violent felonies of “from one to several percent”); D. Michael Risinger,

Innocents Convicted: An Empirically Justified Factual Wrongful Conviction Rate, 97 J. CRIM. L. & CRIMINOLOGY 761, 769–80 (2007) (estimating a wrongful conviction rate of 3.3–5% for

capital rape-murders during the 1980s); George C. Thomas III, Where Have All the Innocents

Gone?, 60 ARIZ. L. REV. 865, 872–79 (2018) (estimating a wrongful conviction rate in North

Carolina of from one-eighth to one-half of one percent). Whatever the merits of these estimates and their underlying methodologies, they tell us nothing about the error rate that inheres in

identification or other particular types of evidence.

203 For a helpful discussion along these lines, see Cassell, supra note 197, at 837–38.

2020] EYEWITNESS IDENTIFICATION 227

If neither the social science research, the Constitution, nor our legal

tradition offers a workable rule for identifying evidence too unreliable to be

admitted, then the determination whether the benefits of more restrictive

identification procedures to protect the innocent are worth the costs that those

procedures may impose in terms of the loss of identifications of the guilty

seems like a policy and not a legal question. Indeed, this is a particularly

difficult policy question given the difficulties not only in quantifying the

costs and benefits of reform but also in assigning costs and benefits to both

the conviction of the innocent and the acquittal of the guilty. 204

In the face of the difficulties of assessing these costs and benefits, it

should be unsurprising that courts have hewed to Manson’s deferential view.

Politically accountable legislatures seem far better positioned to assess these

issues—and to be held accountable for their errors. 205

By excluding evidence

that poses only the most extreme risks to the innocent and relying on the

traditional view that the jury is the appropriate body to assess the probative

value of evidence, Manson may represent about the best we can do. Indeed,

Manson reflects the approach one would expect in the absence of a

justification for prophylaxis—requiring the defendant to demonstrate the

unreliability of the identification on a case-by-case basis, under the totality

of the circumstances.

The preceding discussion rests on the view that the appropriate

objective of policy reform would be to reduce the rate at which factually

innocent defendants are convicted on the basis of inaccurate identification

evidence. 206

As we have seen, it is difficult to justify prophylactic rules

thought to minimize the risk of wrongful conviction without logically

204 For helpful discussions of the difficulty in assessing the costs and benefits of identification procedures, see IDENTIFYING THE CULPRIT, supra note 4, at 76–91; and Clark,

supra note 155, at 246–52.

205 Cf. Medina v. California, 505 U.S. 437, 445–46 (1992) (“[B]ecause the States have

considerable expertise in matters of criminal procedure and the criminal process is grounded

in centuries of common-law tradition, it is appropriate to exercise substantial deference to legislative judgments in this area.”).

206 Perhaps this claim can even be framed in terms of a constitutional right to be free from

factually inaccurate convictions. The Supreme Court has yet to squarely recognize such a

right, although it has been willing to assume that it exists. See, e.g., Dist. Att’y’s Office for Third Judicial Dist. v. Osborne, 557 U.S. 52, 71–72 (2009) (“As a fallback, Osborne also

obliquely relies on an asserted federal constitutional right to be released upon proof of ‘actual

innocence.’ Whether such a federal right exists is an open question. We have struggled with

it over the years, in some cases assuming, arguendo, that it exists.”) (footnote omitted). See generally Jackson v. Virginia, 443 U.S. 307, 324 (1979) (recognizing a due process right to

overturn a conviction on appeal “if it is found that upon the record evidence adduced at the

trial no rational trier of fact could have found proof of guilt beyond a reasonable doubt”).

228 ROSENTHAL [Vol. 110

excluding virtually all evidence of guilt. Perhaps, however, this represents

the wrong perspective on the problem of eyewitness identification.

A different Blackstonian argument is premised not on the desirability of

a prophylactic rule, but rather on the right of all defendants not to have the

prosecution unfairly manipulate the evidence. For example, under what is

perhaps the most Blackstonian aspect of constitutional doctrine, aside from

the burden of proof itself, the Supreme Court has held that the right to a fair

trial under the Due Process Clause is violated when the prosecution

suppresses material exculpatory evidence. 207

From this, one could argue that

when investigators use unnecessarily suggestive procedures, they

compromise the right to a fair trial by manipulating identification evidence

through the use of suggestive procedures, thereby effectively depriving the

defendant of the ability to obtain a potentially exculpatory non-identification

that is not tainted by suggestion. 208

Accordingly, if officials take steps to increase the likelihood that a

witness will make a positive identification through official suggestion, one

could argue that an accused is deprived of his right to a fair trial under the

Due Process Clause. On this view, Manson’s rejection of a per se rule of

exclusion of identifications obtained through unnecessarily suggestive

procedures was a serious error. A per se rule of exclusion would not over-

protect the right in a prophylactic sense, because any effort to manipulate the

evidence is violative of the accused’s rights, regardless of whether a resulting

identification proves to be factually accurate. It is to this contention that we

finally turn.

207 See, e.g., Strickler v. Greene, 527 U.S. 263, 280–81 (1999) (“In Brady [v. Maryland],

this Court held ‘that the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to

punishment, irrespective of the good faith or bad faith of the prosecution.’ We have since held

that the duty to disclose such evidence is applicable even though there has been no request by

the accused, and that the duty encompasses impeachment evidence as well as exculpatory evidence . . . . Moreover, the rule encompasses evidence ‘known only to police investigators

and not to the prosecutor.’” (quoting Brady v. Maryland, 373 U.S. 83, 87 (1967), and Kyles

v. Whitley, 514 U.S. 419, 437 (1995))) (citations omitted).

208 Cf. Rosenberg, supra note 15, at 292–93 (“Conducting an unnecessarily suggestive pretrial identification procedure is analogous to creating one piece of evidence, the

identification that results from the procedure, and destroying another piece of evidence, the

identification, or failure of identification, that would have resulted from a correctly conducted

process. Indeed, an unnecessarily suggestive procedure threatens to compromise all of the subsequent identification testimony by the witness . . . . Given the powerful anchoring effects

of the suggestive procedure on any subsequent identification, as a practical matter a non-

suggestive procedure cannot be conducted after a suggestive one.”) (footnote omitted).

2020] EYEWITNESS IDENTIFICATION 229

B. ASSESSING RELIABILITY IN LIGHT OF THE TOTALITY OF

CIRCUMSTANCES

Some of the criticisms of Manson seem unwarranted. For example,

some argue that Manson’s listing of the factors that bear on the reliability of

an identification is inconsistent with research identifying a variety of other

factors and casting doubt on the importance of some of the listed factors. 209

Yet, on this point, Manson, after observing that “reliability is the linchpin in

determining the admissibility of identification testimony,” added, “[t]he

factors to be considered are set out in [Neil v.] Biggers.” 210

Neil v. Biggers,

in turn, stated that the relevant factors “include the opportunity of the witness

to view the criminal . . . the witness’ degree of attention, the accuracy of his

prior description . . . the level of certainty demonstrated at the confrontation,

and the time between the crime and the confrontation.” 211

Accordingly, Manson and Neil embraced an inclusive approach; they

did not preclude inquiry into any factor with a demonstrable relationship to

reliability, or evidence casting doubt on reliability. Indeed, Manson

repeatedly described the rule it embraced as involving inquiry into “the

totality of the circumstances.” 212

Thus, it should be unsurprising that a

number of lower courts have concluded that Manson does not forbid inquiry

into any factors that bear on reliability, even if not expressly listed in the

Manson opinion, 213

nor does it preclude courts from casting a skeptical eye

on factors listed as relevant in the opinion, but which subsequent research

suggests are of limited significance. 214

On a related issue, however, there is greater uncertainty as to Manson’s

meaning—when assessing the reliability of a suggestive identification, does

Manson permit consideration of independent corroborative evidence, or only

209 See supra text accompanying notes 45–51.

210 Manson v. Brathwaite, 432 U.S. 98, 114 (1977) (citing Neil v. Biggers, 409 U.S. 188,

199–200 (1972)).

211 Neil, 409 U.S. at 199–200 (emphasis supplied).

212 Manson, 432 U.S. at 106, 110, 113.

213 See, e.g., State v. Kaneaiakala, 450 P.3d 761, 777–78 (Haw. 2019) (without rejecting Manson, the court held that lower courts should consider all factors shown to bear on

reliability); State v. Almaraz, 301 P.3d 242, 251–53 (Idaho 2013) (explaining that courts

applying Manson should consider all systems and estimator variables identified in Henderson

when determining whether the use of identification evidence deprives a defendant of due process of law).

214 See, e.g., State v. Discola, 184 A.3d 1177, 1187–89 (Vt. 2018) (rejecting use of a

witness’s degree of certainty as a factor in due process challenges to eyewitness identification

testimony despite its use in Manson in light of subsequent empirical evidence); cf. State v. Dickson, 141 A.3d 810, 824 (Conn. 2016) (“[F]irst time in-court identifications, like in-court

identifications that are tainted by an unduly suggestive out-of-court identification, implicate

due process protections and must be prescreened by the trial court.”).

230 ROSENTHAL [Vol. 110

evidence related to the reliability of the identification itself? Consideration

of this issue, we will see, sheds considerable light on the difficulties of

exclusionary rules for suggestive identifications more robust than Manson.

1. The Role of Corroborative Evidence

A per se rule requiring the exclusion of identifications obtained through

what are regarded as unduly suggestive procedures, of course, would reject

any inquiry into whether the identification is corroborated by other evidence.

In Manson, with respect to the role of independent corroborative evidence,

the Court wrote, “Although it plays no part in our analysis . . . assurance as

to the reliability of the identification is hardly undermined by the facts that

[Brathwaite] was arrested in the very apartment where the sale had taken

place, and that he acknowledged his frequent visits to that apartment.” 215

Thus, in a single sentence, the Court both disclaimed reliance on

corroborative evidence and suggested that corroboration bears on reliability.

In contrast, in his separate opinion, Justice Stevens wrote: “[I]t is sometimes

difficult to put other evidence of guilt entirely to one side . . . . [T]he Court

carefully avoids this pitfall and correctly relies only on appropriate indicia of

the reliability of the identification itself.” 216

Yet, this seems an

overstatement; the opinion of the Court, while treating the corroborative

evidence as unnecessary to the outcome in that case, seemed to acknowledge

that corroboration bears on reliability. It is difficult to read Manson as

containing a square holding that corroborative evidence is irrelevant when

assessing reliability under the totality-of-the-circumstances test. 217

Indeed, when it comes to the role of corroborative evidence, the lower

courts have split. Some have concluded that Manson permits consideration

of only evidence relating to reliability of the identification itself, with

corroborative evidence independent of the identification relevant only on the

question whether the erroneous admission of identification evidence amounts

to harmless error. 218

Others treat independent evidence that corroborates the

215 Manson, 432 U.S. at 116 (1977) (footnote omitted).

216 Id. at 118 (Stevens, J., concurring) (footnote omitted).

217 Cf. GARRETT, supra note 15, at 81–85 (discussing contrasting understandings of

Manson’s application to corroborative evidence).

218 See, e.g., United States v. Greene, 704 F.3d 284, 308–10 (4th Cir. 2012); Raheem v.

Kelly, 257 F.3d 122, 140–41 (2d Cir. 2001); United States v. Rogers, 126 F.3d 655, 659–60

(5th Cir. 1997); United States v. Emanuele, 51 F.3d 1123, 1128 (3d Cir. 1995); Green v.

Loggins, 614 F.2d 219, 224–25 (9th Cir. 1980); Long v. United States, 156 A.3d 698, 707–08 (D.C. 2017); State v. Jones, 128 A.3d 1096, 1107–08 (N.J. 2016); Wise v. Commonwealth,

367 S.E.2d 197, 201–02 (Va. Ct. App. 1988); Campbell v. State, 589 P.2d 358, 364–65 (Wyo.

1979). For endorsements of this view, see Rosenberg, supra note 15, at 286–88; and Rudolf

2020] EYEWITNESS IDENTIFICATION 231

reliability of a contested identification as bearing on its admissibility under

Manson. 219

There are, to be sure, reasons to resist the use of corroborative evidence

when assessing the reliability of an identification. For one thing,

corroborative evidence could distort the reliability inquiry by bootstrapping

the reliability of an identification to other evidence. 220

For another,

unnecessarily suggestive identification procedures, one could argue, advance

no legitimate governmental interest, but instead degrade the reliability of the

criminal process. 221

Beyond that, a per se exclusionary rule would

incentivize investigators to utilize procedures that minimize the risk of

Koch, Note, Process v. Outcome: The Proper Role of Corroborative Evidence in Due Process

Analysis of Eyewitness Identification Testimony, 88 CORNELL L. REV. 1097, 1123–40 (2003).

219 See, e.g., United States v. Constant, 814 F.3d 570, 575–77 (1st Cir. 2016) (identification corroborated by defendant’s confession and fit between description of shooter’s

gun and his weapon); United States v. Rogers, 73 F.3d 774, 778 (8th Cir. 1996) (independent

witnesses corroborated contested identification); United States ex rel. Kosik v. Napoli, 814

F.2d 1151, 1156–57, 1161 (7th Cir. 1987) (identification corroborated by independent witnesses’ description of vehicle containing suspects); Graham v. Solem, 728 F.2d 1533,

1546–48 (8th Cir. 1984) (identification corroborated when defendants admitted to being

together at the time and scene of the crime); McNary v. Sowders, 660 F.2d 703, 708–09 (6th

Cir. 1981) (identification corroborated by recovery of pistol taken in robbery and thrown from vehicle in which defendant was riding); Lindsey v. State, 572 S.W.2d 145, 146 (Ark. 1978)

(identification corroborated by recovery from defendant of cash in denominations identical to

what was taken in robbery); People v. Lee, 502 N.E.2d 399, 407–08 (Ill. Ct. App. 1986)

(identification corroborated by defendant’s confession); St. Clair v. Commonwealth, 140 S.W.3d 510, 551–52 (Ky. 2004) (identification corroborated by independent testimony,

forensic and circumstantial evidence); State v. Egana, 792 So. 2d 931, 935–37 (La. Ct. App.

2001) (identification corroborated by videotape); Commonwealth v. Hicks, 460 N.E.2d 1053,

1055–60 (Mass. Ct. App. 1984) (identification corroborated by recovery of fruits of robbery from defendant shortly after crime), overruled on other grounds by Commonwealth v.

Johnson, 650 N.E.2d 1257 (Mass. 1995); Johnson v. State, 354 P.3d 667, 676 (Nev. Ct. App.

2015) (identification corroborated when victim’s cellphone was found in pocket on

accomplice); State v. Halley, 637 N.E.2d 937, 941 (Ohio Ct. App. 1994) (identification corroborated because suspect was only male staying at the apartment where attack occurred

and was found in clothes described by victim); State v. Brown, 589 S.E.2d 781, 785–87 (S.C.

Ct. App. 2003) (identification corroborated when defendant was found where witness stated

he would be)

220 See, e.g., Koch, supra note 218, at 1134 (“[I]f courts were to consider evidence of general guilt in determining whether an identification is reliable, that evidence would

essentially be counted twice—first toward general guilt, then again toward admitting the

identification, which would, in turn, act as further evidence of guilt. This evidence would

therefore be weighted too heavily, to the point that outcomes could become distorted.”).

221 See, e.g., Rosenberg, supra note 15, at 291 (“[A]n unnecessarily suggestive

identification procedure simply creates unreliable evidence where reliable evidence could

have been gathered.”) (footnote omitted).

232 ROSENTHAL [Vol. 110

suggestion more effectively than a rule that tolerates suggestion as long as

corroborative evidence is obtained. 222

The arguments against the use of corroborative evidence are perhaps

best considered by reference to particular facts, rather than in the abstract.

To that end, consider once more the facts of Henderson. The extent to which

Womble’s identification of Henderson was corroborated was striking—not

only did Clark (the shooter) also place Henderson at the apartment, 223

but

Henderson himself admitted that he was there. 224

It is difficult to conclude,

in light of this evidence, that Womble’s identification was unreliable in the

sense that it was likely to lead the jury into error.

Consider as well the question whether Detective Ruiz and Investigator

MacNair engaged in anything that should be characterized as unnecessarily

suggestive. Even if they put some pressure on Womble to make an

identification, recall that Womble had previously lied to them about his

knowledge of the shooting, and later admitted that he was afraid to make an

identification because he had been threatened. 225

In high-crime communities,

the threat of retaliation is often quite real; for example, the pertinent literature

reflects the prevalence of intimidation tactics by urban street gangs as a

means of inhibiting community cooperation with the police. 226

Indeed,

researchers have found that “offenders in gang-related and drug-related

homicides are much less likely to be arrested . . . in part due to lack of witness

cooperation.” 227

In light of this, when a witness who has previously admitted

to being in proximity to a homicide, and who has expressed a fear of

retaliation that has already compromised his candor with the authorities, then

fails to make an identification during an ensuing identification procedure, it

222 See, e.g., Manson v. Brathwaite, 432 U.S. 98, 125 (1977) (Marshall, J., dissenting)

(“[T]he Court acknowledges that one of the factors, deterrence of police use of unnecessarily

suggestive identification procedures, favors the per se rule [requiring exclusion of

identification obtained by unnecessarily suggestive procedures]. Indeed, it does so heavily, for such a rule would make it unquestionably clear to the police they must never use a suggestive

procedure when a fairer alternative is available.”).

223 At trial, the court admitted Clark’s videotaped statement which, while placing

Henderson in the apartment at the time of the shooting, largely exculpated him. Id. at 882 &

n.2.

224 Id. at 880, 882.

225 See supra text accompanying notes 87–89.

226 See, e.g., AL VALDEZ, UNDERSTANDING STREET GANGS 19–20 (3rd ed. 1997); Bruce

D. Johnson et al., Drug Abuse in the Inner City: Impact on Hard-Drug Users and the

Community, in CRIME AND JUSTICE.: A REVIEW OF RESEARCH 9, 35–37 (Michael Tonry &

James Q. Wilson eds., 1990).

227 Anthony A. Braga & Rod K. Brunson, The Police and Public Discourse on “Black-

On-Black” Violence, in NEW PERSPECTIVES ON POLICING 2015, at 6 (Nat’l Inst. Of Just., Pub.

No. 248588, May 2015).

2020] EYEWITNESS IDENTIFICATION 233

is far from clear that it is inappropriate for investigators to confront the

witness rather than simply giving up, especially when a jury will

subsequently be able to assess all of the circumstances surrounding the

identification, including the witness’s initial reluctance.

The facts of Henderson suggest two important, if interrelated,

difficulties with a per se rule requiring suppression of all identification

evidence that results from unnecessarily suggestive procedures.

First, it will not always be obvious what should be characterized as

unnecessary suggestion or an improper effort to tamper with identification

evidence. In a great many cases, it will not be difficult for the defense to

identify some additional precaution that could have been taken to avoid

suggestion. Thus, defense counsel will likely be able to characterize many

identifications as unnecessarily suggestive. Yet, determining whether some

additional precaution was appropriate will often be challenging. Even

though Detective Ruiz and Investigator MacNair compromised the double-

blind protocol, it could well have been necessary for them to confront

Womble to overcome his fear of retaliation. We cannot be confident whether

the actions taken by the officers were necessary or not. Perhaps Womble was

genuinely uncertain of his identification and yielded to police pressure; or

perhaps Womble was unwilling to become the chief prosecution witness in a

murder case in the face of a threat of retaliation. We cannot know for sure.

The data set out above, however, suggests that more rigorous identification

protocols will sometimes result in the loss of accurate identifications.

Perhaps procedures that some might characterize as unnecessarily

suggestive, others could justifiably believe are necessary to obtain probative

evidence.

A rule requiring exclusion of unnecessarily suggestive identifications

could be thought justifiable to the extent that it deters official misconduct. 228

It will frequently be difficult, however, to determine if an officer’s response

to a witness’s failure to make an identification represents misconduct or a

necessary prod to a reluctant or fearful witness. If a finding of unnecessary

suggestion resulted in a rule of automatic exclusion, quite high stakes would

be placed on the resolution of what, in the real world, is the difficult question

of whether any particular tactic was necessary to obtain useful information

from fearful witnesses.

228 Cf. Katherine R. Kruse, Wrongful Convictions and Upstream Reform in the Criminal

Justice System, 3 TEX. A&M L. REV. 367, 382 (2015) (“As upstream reforms to eyewitness identification procedures gain momentum, the failure of law enforcement agencies to adopt,

enforce, or follow them—especially in the face of statutory mandates—could be seen as a

form of deliberate misconduct warranting a deterrent sanction.”).

234 ROSENTHAL [Vol. 110

The point can be made in doctrinal terms. While the suppression of

material exculpatory evidence violates the Due Process Clause, the use of

what are regarded as suggestive identification procedures presents a different

problem. Because we cannot know whether the witness would have made an

identification had different identification protocols been used, the use of

suggestive identification procedures deprives the accused of only potentially

exculpatory evidence—we cannot know for sure, because the inquiry is

counterfactual. The Supreme Court has held that when officials destroy or

otherwise make unavailable potentially exculpatory evidence, due process is

violated only if officials have acted in bad faith. 229

This is, accordingly, the

doctrinal framework applicable to a failure to utilize more rigorous

identification protocols. As we have seen, however, more rigorous

identification protocols that minimize the risk of suggestion may also make

it harder to obtain accurate identifications. 230

A failure to utilize prophylactic

procedures may reflect bad faith, but it also could reflect an effort to avoid

the unnecessary loss of important evidence.

In Henderson, for example, it is far from clear that Detective Ruiz and

Investigator MacNair acted in bad faith, even if they undermined the

prescribed double-blind identification protocol. Perhaps they had accurately

perceived Womble’s reluctance to become a prosecution witness in a

homicide case and compromised the double-blind protocol only for that

reason. To be sure, if courts announced a per se rule requiring exclusion of

any identification obtained after blind administration had been compromised,

then the detectives’ conduct could be characterized as a bad-faith violation

of a known legal duty, but one cannot justify such a rule merely by assuming

that officers act in bad faith when they compromise blind administration in

the absence of a per se rule demanding blind administration. Sometimes,

officers may conclude that blind administration has impeded their ability to

confront a recalcitrant witness who balks at making an accurate

identification. If so, perhaps a decision to confront a potentially recalcitrant

witness may have a fully sufficient law enforcement justification.

Second, and relatedly, Henderson also demonstrates that sometimes

there will be little reason to doubt the accuracy of an identification even in

the face of official suggestion. After all, Womble’s identification was

corroborated by both Clark and Henderson himself. Even if an identification

made under suggestive circumstances might then induce police to seek

corroborative evidence, this does not inevitably mean that the resulting

corroboration is unreliable. There is no indication in Henderson, for

229 See Arizona v. Youngblood, 488 U.S. 51, 57–58 (1988).

230 See supra Part II.A.1.

2020] EYEWITNESS IDENTIFICATION 235

example, that Clark’s identification of Henderson, or Henderson’s admission

that he was present at the scene of the shooting, represents mere

bootstrapping onto a suspect identification. 231

Even if a potentially

suggestive identification causes the police to seek corroborative evidence,

the resulting corroboration may well demonstrate the accuracy of the

identification. The existence of corroborative evidence does not negate the

fact that a suggestive identification procedure was used, but it can

demonstrate that whatever the suggestion, it poses little risk of convicting an

innocent suspect.

The point is not unique to Henderson. In Stovall v. Denno, for example,

although Stovall was identified by the victim when he was brought to her

hospital room in police custody under highly suggestive circumstances,

police found a shirt at the scene of the attack containing keys that they traced

to Stovall. 232

In Perry v. New Hampshire, Perry’s contested identification as

the individual who had been trying to break into parked cars occurred after

he had been taken into police custody by an officer who found Perry standing

between parked cars and holding two car amplifiers with a metal bat lying on

the ground behind him, and then learned that the rear window of a nearby car

was shattered and its speakers and amplifiers were missing. 233

In

Commonwealth v. Hicks, although the victim made an identification only

when police told him that they had apprehended the men who had robbed

him, the suspects had been found near the scene of the crime and shortly after

it occurred, in possession of the items taken in the robbery. 234

In each of

these cases, the identification procedures were suggestive, but there was also

little doubt that the identifications were accurate. Indeed, we have seen

numerous cases in which corroborative evidence convincingly demonstrated

the reliability of an identification obtained through what were likely

unnecessarily suggestive procedures. 235

Whatever manipulation may have

occurred in the identification process, its outcome in these cases was reliable.

While there may be some cases in which an identification is mistaken despite

corroboration, there are surely many others in which corroborative evidence

will greatly reduce the likelihood that an identification was erroneous.

As it happens, one of the reforms that some critics have advocated to

protect the innocent is a requirement that convictions that rest on eyewitness

231 See supra text accompanying notes 87–90, 183–184.

232 See Stovall v. Denno, 388 U.S. 293, 295 (1967).

233 Perry v. New Hampshire, 565 U.S. 228, 233–34 (2012).

234 Commonwealth v. Hicks, 460 N.E.2d 1053, 1054–55 (Mass. Ct. App. 1984), overruled

on other grounds by Commonwealth v. Johnson, 650 N.E.2d 1257 (Mass. 1995).

235 See supra text accompanying notes 183–190.

236 ROSENTHAL [Vol. 110

identification testimony not be sustained absent corroboration. 236

This

position is based on the insight that corroboration is one way in which

evidence can be shown to be reliable. Indeed, many types of evidence—the

testimony of an accomplice offered leniency, or a jailhouse informant, or a

confession exacted in the face of aggressive interrogation techniques—might

be regarded as unreliable in the abstract, but once corroborated, can constitute

powerful—and reliable—evidence of guilt.

2. Corroborated Identifications and the Right to a Fair Trial

As we have seen, it is difficult to assess the reliability of any piece of

evidence in the abstract; it is frequently only when the totality of the evidence

is assessed that it becomes possible to reach judgments about reliability. 237

It is not obvious why a defendant is denied a fair trial when convicted on the

basis of evidence that proves reliable in light of the available corroboration.

Even if corroborative evidence is used twice during a prosecution—

once to convince the judge of the admissibility of an identification under

Manson, and again to convince the trier of fact of the defendant’s guilt—that

does not mean that the resulting conviction is not supported by appropriate

proof. Corroborative evidence is often doubly considered in this fashion—

for example, proffered hearsay is considered, together with corroborative

evidence, both to determine its admissibility, and, subsequently, as

substantive proof of guilt at trial. 238

Nor does the use of evidence in this

fashion involve a logical bootstrap. If the reliability of evidence is to be

considered both when it is admitted and again when determining whether the

defendant’s guilt has been proven, then corroborative evidence bearing on

reliability is logically relevant at both stages.

236 See, e.g., Kruse, supra note 228, at 388–92; Boaz Sangero & Mordechai Halpert, Why

A Conviction Should Not Be Based on a Single Piece of Evidence: A Proposal for Reform, 48

JURIMETRICS J. 43, 90–94 (2007); Sandra Guerra Thompson, Beyond a Reasonable Doubt? Reconsidering Uncorroborated Eyewitness Identification Testimony, 41 U.C. DAVIS L. REV.

1487, 1523–43 (2008); cf. Wells & Quinlivan, supra note 6, at 20 (arguing that when

suggestive procedures are used the burden should be placed on the prosecution to demonstrate

reliability by evidence independent of the identification). There are, however, a great many difficulties in fashioning and administering a standard for corroboration. For an illustrative

discussion, see David Crump, Eyewitness Corroboration Requirements As Protection Against

Wrongful Conviction: The Hidden Questions, 7 OHIO ST. J. CRIM. L. 361 (2009).

237 See supra text accompanying note 182.

238 See, e.g., Bourjaily v. United States, 483 U.S. 171, 179–81 (1987) (explaining that although hearsay is presumed unreliable, an alleged coconspirator’s hearsay statements may

be considered together with other evidence to determine both admissibility as the statement of

a conspirator and as proof of guilt at trial).

2020] EYEWITNESS IDENTIFICATION 237

To be sure, the courts that ignore independent corroborative evidence

when assessing the admissibility of an identification still consider such

evidence on the question whether the improper admission of identification

evidence was harmless error. 239

Perhaps, in the examples canvassed above,

the identifications should have been excluded, yet the convictions could have

been upheld on the ground that the error in admitting the identifications was

harmless in light of the corroborative evidence.

Nevertheless, even if erroneously admitted identification evidence is

sometimes harmless error, an exclusionary rule more robust than Manson, or

one that required the use of demanding identification protocols that reduce

the rate at which even accurate identifications are made, could still have

dramatic effects, even in cases with powerful corroborative evidence. After

all, just as “the in-court testimony of an eyewitness can be devastatingly

persuasive,” 240

an eyewitness’s inability to provide such testimony might be

of great significance as well, even in light of corroborative evidence. If the

key witness is not able to identify the defendant, skillful defense counsel

could persuasively argue that the lack of an identification by the key

eyewitness suggests reasonable doubt, despite the remaining evidence.

Indeed, some jurors may draw an adverse inference from a key witness’s

failure to identify the defendant—affording it perhaps greater weight than is

warranted. 241

Under a regime in which trial courts exclude all suggestive

identifications, conviction rates may well decline if jurors are troubled by the

failure of the prosecution to adduce identification testimony from key

witnesses.

239 See supra text accompanying note 214. The use of unconstitutionally obtained

identification evidence is considered harmless if it is shown beyond reasonable doubt that the improper evidence had no effect on the verdict. See, e.g., Moore v. Illinois, 434 U.S. 220, 232

(1977) (citing Chapman v. California, 386 U.S. 18 (1967)).

240 United States v. Greene, 591 F.2d 471, 475 (8th Cir. 1979).

241 This problem arises both in cases in which witnesses are unable to make accurate

identifications because of the use of more rigorous protocols, and in cases in which jurors are unaware that the witness actually made an identification because it was excluded from

evidence prior to trial. In the latter cases, the problem is exacerbated because when an

identification is excluded, all subsequent identifications must also be excluded unless based

on an independent source, which is a difficult showing to make; see, e.g., Thompson, supra note 15, at 627 (“Based on scientific studies about memory distortion, a strong argument can

be made that an earlier suggestive identification procedure will permanently distort any

later identification by the same witness, including an in-court identification. Thus,

a suggestive pretrial identification procedure renders any in-court identification just as unreliable as the pretrial identification.”) (footnote omitted). For this reason, a per se rule

excluding all unnecessarily suggestive identifications is likely to taint any subsequent effort

to obtain an identification from the witness.

238 ROSENTHAL [Vol. 110

Most important, however, is the question whether the use of a

corroborated identification should be regarded as constitutional error in the

first place. As we have seen, in terms of existing doctrine, governmental

conduct that deprives a defendant of potentially exculpatory evidence

violates due process only if undertaken in bad faith, and it is difficult to

characterize the government’s refusal to utilize identification protocols that

make it more difficult for witnesses to make even accurate identifications as

reflecting bad faith. 242

Even putting this point aside, it is difficult to

understand why the use of a reliable identification—whether because it has

been corroborated or for any other reason—somehow deprives a defendant

of a fair trial.

For example, when the fillers in a lineup do not closely resemble the

perpetrator, it may be difficult for a court or a jury to tell whether a resulting

identification is tainted by suggestion. Even an identification based on a

relative judgment that the suspect resembles the perpetrator, however, can

represent probative evidence of guilt. 243

Moreover, when a witness’s relative

judgment about the similarity between the perpetrator and a suspect is

corroborated in a meaningful way, it is hard to deny that even a relative

judgment represents appropriate evidence of guilt, just as eyewitnesses who

never see a perpetrator’s face can be permitted to describe his height, or hair

color. Again, there may be some cases in which even corroborated

identifications prove mistaken, but surely substantial corroboration greatly

reduces the risk of error. For this reason, it becomes difficult to conclude

that the prosecution has deprived the defendant of exculpatory evidence

merely because more rigorous identification protocols were not employed.

The defense is free to argue that had different identification procedures been

used, the result might have been exculpatory, but in light of the available

corroboration, the trier of fact is surely entitled to credit the identification

despite the use of potentially suggestive procedures. At a minimum,

242 See supra text accompanying notes 224–226.

243 Cf. Clark et al., supra note 155, at 66–67:

[T]he absolute-relative distinction is imprecise and may fail at both ends. At one end, a pure

relative model makes a clearly false prediction: if a witness to an armed robbery (by any white

male) were presented with a lineup consisting of George W. Bush and five penguins, he or she

would identify the 43rd President with a high level of confidence. One may reasonably condemn

as illegitimate all identifications based on such a pure relative judgment decision rule, but this

might only provide guidance to condemn a decision strategy that almost no witnesses ever use. At

the other end, a pure version of the absolute judgment strategy, with no relative component, may

also fail as a psychological theory. With few extremely rare exceptions (i.e., perfect pitch),

almost all human judgments involve relative judgments to some degree. Thus, it would not make

sense to declare an identification to be illegitimate simply because it was based in part on relative

judgments, as such a standard would render all eyewitness identifications to be illegitimate.

2020] EYEWITNESS IDENTIFICATION 239

whatever the value of speculation that had more rigorous identification

protocols been used, the defendant would not have been identified,

counterfactual speculation of this character is far different from the

prosecution’s failure to disclose actual exculpatory information of which it is

aware.

Of course, there is an argument for evidentiary rules that deter the use

of unnecessarily suggestive identification procedures. A constitutional

argument for deterrence, however, must be anchored in the accused’s right

to a fair trial, and not merely in an interest in punishing officials for their

failure to utilize what might be regarded as best practices for identification. 244

After all, as we have seen in Part II.A.2 above, the purely prophylactic case

for more rigorous identification protocols is a problematic one. Yet,

establishing that a failure to utilize practices that minimize the risk of

suggestion deprives an accused of the right to a fair trial is problematic.

As we have seen, assessing what constitutes unnecessary suggestion is

a complicated business; the use of less rigorous identification protocols may

be necessary to avoid the loss of even accurate identifications when more

rigorous protocols make identifications more difficult to make. 245

Exclusion

may deter misconduct, but it might also, in a case like Henderson, deter the

police from doing what is necessary to convince a fearful witness to identify

a violent offender. In any event, if corroborative evidence demonstrates that,

despite the suggestiveness of the procedures employed, the identification is

reliable, then the likelihood that suggestive identification procedures

deprived the defendant of exculpatory evidence or otherwise facilitated the

conviction of an innocent person is surely low.

A defendant convicted on the basis of reliable identification evidence—

because it has been corroborated—will accordingly have difficulty

demonstrating that he was unfairly deprived of exculpatory evidence by the

use of potentially suggestive identification procedures. Instead, when the

corroborative evidence is sufficiently powerful, it is far more likely that the

use of more rigorous identification protocols would have made no difference,

or, at most, would have prevented the prosecution from obtaining a reliable

identification of a guilty perpetrator because, as we have seen in Part II.A.1

above, more rigorous identification protocols reduce the rate at which even

accurate identifications are made. 246

244 Cf., e.g., Brady v. Maryland, 373 U.S. 83, 87 (1963) (“The principle . . . is not

punishment of society for misdeeds of a prosecutor but avoidance of an unfair trial to the accused.”).

245 See supra Part II.A.1.

246 For this reason, anchoring a constitutional argument against the use of evidence

obtained through unnecessarily suggestive procedures, but also corroborated by independent

240 ROSENTHAL [Vol. 110

3. The Difficulty of Abandoning Manson

The argument advanced above not only supports the use of

corroborative evidence in applying Manson, but also demonstrates the

difficulties of any Blackstonian approach to the admissibility of eyewitness

identification evidence.

Once one departs from a test of reliability under the totality of

circumstances, one necessarily enters the realm of prophylaxis. After all, a

totality-of-the-circumstances test demands that a judge make the best

assessment possible about the reliability of an identification in light of all

available information. It is perhaps for this reason that even the states that

purport to employ robust exclusionary rules seem to balk at hewing to those

rules, as we have seen. 247

The prosecution’s burden of proof beyond a reasonable doubt already

builds into the criminal process significant protection against convicting the

innocent. We proceed further at our peril. As we have seen, there is no

reliable way to gauge the costs and benefits of a prophylactic rule that would

exclude evidence judges find to be reliable, even if produced by

unnecessarily suggestive identification procedures. Indeed, we have also

seen that it is even difficult to decide what is unnecessarily suggestive, given

that more rigorous identification protocols seem not only to reduce the risk

of false identifications, but also the rate at which guilty perpetrators are

evidence, in the government’s obligation not to suppress exculpatory evidence faces an

additional doctrinal obstacle beyond the defendant’s likely inability to establish bad faith. The

suppression of even actual (not merely potentially) exculpatory evidence does not deprive a defendant of the constitutional right to a fair trial unless there is a reasonable probability that

the suppressed evidence would have produced a different verdict. See, e.g., Strickler v.

Greene, 527 U.S. 263, 289–90 (1999) (To obtain relief, a defendant “must convince us that

‘there is a reasonable probability’ that the result of the trial would have been different if the suppressed documents had been disclosed to the defense. As we stressed in Kyles: ‘[T]he

adjective is important. 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.’” (citation omitted and brackets in original) (quoting Kyles v. Whitley, 514 U.S. 419, 434 (1995)). When there is

adequate corroboration for an identification, it will be difficult for an accused to establish a

reasonable likelihood that the use of an alternative identification procedure would have

exculpated her.

247 See supra Part I.C; cf. Manson, 432 U.S. 98, 112–13 (“[T]he per se approach suffers

serious drawbacks. Since it denies the trier reliable evidence, it may result, on occasion, in the

guilty going free. Also, because of its rigidity, the per se approach may make error by the trial

judge more likely than the totality approach. And in those cases in which the admission of identification evidence is error under the per se approach but not under the totality approach

—cases in which the identification is reliable despite an unnecessarily suggestive

identification procedure—reversal is a Draconian sanction.”) (footnote omitted).

2020] EYEWITNESS IDENTIFICATION 241

correctly identified. 248

Perhaps it is necessary to run some risk of false

identification of the innocent to achieve an acceptable rate at which the guilty

can be identified.

Nor is it easy to explain why due process prevents a conviction obtained

by evidence that a judge finds reliable under the totality of the circumstances,

even if it is possible that more rigorous identification protocols might not

have generated the same evidence. After all, it is hard to know whether the

different result that a more rigorous protocol might produce would reflect a

false identification of an innocent, or a false exoneration of the guilty. In any

event, any effort to assess reliability exclusively by reference to the

procedures used to obtain an identification is a fool’s errand. As we have

seen, the reliability of most evidence cannot be assessed in a vacuum;

reliability can be properly assessed only in light of the totality of the

circumstances. 249

That is what makes Manson preferable to its prophylactic

alternatives. A Blackstonian approach that resists any evidence thought to

raise a risk of wrongful conviction, as we have seen, is one that could bar the

use of pretty much any evidence of guilt. Without a stopping point,

unvarnished Blackstonism becomes an engine for the destruction of the

criminal justice system.

CONCLUSION

It is strong medicine for a court to bar an eyewitness—perhaps even a

victim—from testifying that she sees the perpetrator sitting in the

courtroom. 250

For that reason, it may be that Manson will inevitably be

applied in a deferential manner; in all but the clearest cases, most judges are

likely to balk at the idea of preventing an eyewitness from telling a jury what

she saw. Perhaps expert testimony and jury instructions can sensitize juries

to the risks of eyewitness identification evidence, although, as we have seen,

there is reason to doubt that this will prove anything close to a panacea. 251

Even so, there is no good alternative.

Blackstonians could drive the risk of wrongful conviction on the basis

of eyewitness testimony to zero by barring all of it; but few would regard that

as an acceptable tradeoff. Surely Manson was correct to observe that our

248 See supra Part II.A.1.

249 See supra text accompanying note 182.

250 Cf. Richard A. Rosen, Reflections on Innocence, 2006 WIS. L. REV. 237, 251

(“[I]magine looking a rape victim in the eye, one who swears that she can identify the man

who violated her, and telling that woman she will not even be allowed to tell her story to a jury. It is no wonder that few identifications have been suppressed for due process

violations.”) (footnote omitted).

251 See supra text accompanying notes 168–170.

242 ROSENTHAL [Vol. 110

conception of due process includes concern about preserving the ability of

the prosecution to have a fair opportunity to convict the guilty. 252

Absent far

clearer empirical evidence about the costs and benefits of prophylaxis than

can be found in current research, Manson’s totality-of-the-evidence

approach, with all its imperfections, is likely the best we can do.

As long as fallible people are involved in the administration of justice,

there will be a risk of wrongful conviction. To be sure, there are ample

reasons to endeavor to minimize the risk of wrongful conviction. It is no

small feat, however, to design Blackstonian reforms that drive down the risk

of wrongful conviction of the innocent without increasing the rate at which

the guilty go free.

If the guilty must go unpunished to vindicate a constitutional limitation

on the ability of the government to obtain evidence—such as the Fourth

Amendment’s prohibition on unreasonable search and seizure—then perhaps

that is a price that the Constitution itself exacts. 253

No such constitutional

limitation is at stake, however, when the objection to evidence obtained by

the government is that it may be unreliable. If reliability is the constitutional

concern, then a totality-of-the-circumstances test for reliability seems well-

suited to address the problem. It is far from evident that a prophylactic

approach would be superior. A totality-of-the-circumstances test will

produce an error rate—especially given the reluctance of courts to exclude

eyewitness accounts—but prophylaxis, by its nature, produces an error rate

as well.

Perhaps Manson is doomed to be applied in a manner that renders it

largely toothless except in the face of the clearest examples of official

suggestion, but that may well be the best that we can expect when the

judiciary must grapple with the difficult tradeoff between the risk of

convicting the innocent if eyewitness testimony is wrong, and acquitting the

guilty if the eyewitness testimony kept from the trier of fact. Perhaps due

252 See Manson, 432 U.S. at 111–13 (“There are, of course, several interests to be

considered and taken into account,” including “the effect on the administration of justice,” and adding that “inflexible rules of exclusion that may frustrate rather than promote justice have

not been viewed recently by this Court with unlimited enthusiasm.”).

253 Cf. Potter Stewart, The Road to Mapp v. Ohio and Beyond: The Origins, Development

and Future of the Exclusionary Rule in Search-and-Seizure Cases, 83 COLUM. L. REV. 1365, 1392–93 (1983) (“Much of the criticism leveled at the exclusionary rule is misdirected; it is

more properly directed at the fourth amendment itself. It is true that, as many observers have

charged, the effect of the rule is to deprive the courts of extremely relevant, often direct

evidence of the guilt of the defendant. But these same critics sometimes fail to acknowledge that, in many instances, the same extremely relevant evidence would not have been obtained

had the police officer complied with the commands of the fourth amendment in the first

place.”) (footnotes omitted).

2020] EYEWITNESS IDENTIFICATION 243

process must accommodate the reality that achieving an optimal balance

between those two risks is a hopeless task. The jury may be an imperfect

vehicle for assessing eyewitness evidence, but it is the vehicle for resolving

guilt or innocence found in the Constitution. We can have little confidence

that a judge-made substitute will do better.

Copyright of Journal of Criminal Law & Criminology is the property of Northwestern University School of Law and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.